Recovery of menses after functional hypothalamic amenorrhoea: if, when and why

Abstract

BACKGROUND

Prolonged amenorrhoea occurs as a consequence of functional hypothalamic amenorrhoea (FHA) which is most often induced by weight loss, vigorous exercise or emotional stress. Unfortunately, removal of these triggers does not always result in the return of menses. The prevalence and conditions underlying the timing of return of menses vary strongly and some women report amenorrhoea several years after having achieved and maintained normal weight and/or energy balance. A better understanding of these factors would also allow improved counselling in the context of infertility. Although BMI, percentage body fat and hormonal parameters are known to be involved in the initiation of the menstrual cycle, their role in the physiology of return of menses is currently poorly understood. We summarise here the current knowledge on the epidemiology and physiology of return of menses.

OBJECTIVE AND RATIONALE

The aim of this review was to provide an overview of (i) factors determining the recovery of menses and its timing, (ii) how such factors may exert their physiological effects and (iii) whether there are useful therapeutic options to induce recovery.

SEARCH METHODS

We searched articles published in English, French or German language containing keywords related to return of menses after FHA published in PubMed between 1966 and February 2020. Manuscripts reporting data on either the epidemiology or the physiology of recovery of menses were included and bibliographies were reviewed for further relevant literature. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria served to assess quality of observational studies.

OUTCOMES

Few studies investigate return of menses and most of them have serious qualitative and methodological limitations. These include (i) the lack of precise definitions for FHA or resumption of menses, (ii) the use of short observation periods with unsatisfactory descriptions and (iii) the inclusion of poorly characterised small study groups. The comparison of studies is further hampered by very inhomogeneous study designs. Consequently, the exact prevalence of resumption of menses after FHA is unknown. Also, the timepoint of return of menses varies strongly and reliable prediction models are lacking. While weight, body fat and energy availability are associated with the return of menses, psychological factors also have a strong impact on the menstrual cycle and on behaviour known to increase the risk of FHA. Drug therapies with metreleptin or naltrexone might represent further opportunities to increase the chances of return of menses, but these require further evaluation.

WIDER IMPLICATIONS

Although knowledge on the physiology of return of menses is presently rudimentary, the available data indicate the importance of BMI/weight (gain), energy balance and mental health. The physiological processes and genetics underlying the impact of these factors on the return of menses require further research. Larger prospective studies are necessary to identify clinical parameters for accurate prediction of return of menses as well as reliable therapeutic options.

recovery of menses, return of menses, restoration of the menstrual cycle, weight, BMI, percentage body weight, eating disorders, physical exercise, stress

Introduction

The human menstrual cycle is essential not only for reproduction but also for general well-being (Walf and Frye, 2006). Oestradiol is needed for adequate bone, cardiovascular, mental and vaginal health (Manonai et al., 2004; Rettberg et al., 2014; Filova et al., 2015; Gordon et al., 2017; Iorga et al., 2017; Levin et al., 2018). Weight loss, eating disorders, exercise and emotional stressors suppress the activity of the GnRH neuronal network and can lead to functional hypothalamic amenorrhoea (FHA) (Drew, 1961; Fries et al., 1974; Frisch and McArthur, 1974; Mecklenburg et al., 1974; Bullen et al., 1985; Gadpaille et al., 1987; Berga and Girton, 1989; Loucks et al., 1989; Pirke et al., 1989; Warren et al., 1999; Roa et al., 2010; Berga and Naftolin, 2012; Garcia-Garcia, 2012; Sanchez-Garrido and Tena-Sempere, 2013; Castellano and Tena-Sempere, 2016). A reduction in the GnRH drive results in abnormal LH pulse frequency generating cycle disturbances (Berga and Girton, 1989; Laughlin et al., 1998; Ackerman et al., 2012). By definition, FHA is only diagnosed after anatomic or organic causes of amenorrhoea have been excluded.

FHA is responsible for 20–35% of secondary amenorrhoea (Practice Committee of American Society for Reproductive Medicine, 2008). It occurs in up to 89% of women with anorexia nervosa and up to 60% of high-performance athletes (Sanborn et al., 1982; Warren and Perlroth, 2001; Watson and Andersen, 2003; Andersen and Ryan, 2009; Roupas and Georgopoulos, 2011). Because of pre-existing cycle irregularities in some women, the concept of post-pill amenorrhoea was abandoned in the early 1980s (Jacobs et al., 1977; Barnhart and Schreiber, 2009); 97% of women experience return of menses within a median time of 32 days after discontinuation of combined oral contraceptives (Davis et al., 2008).

FHA is currently treated symptomatically by hormonal replacement therapy to prevent unfavourable health consequences resulting from a lack of oestrogens. However, hormone therapy may be associated with an increased risk of breast cancer, venous thromboembolism or stroke (Magliano et al., 2006; Marjoribanks et al., 2017; Collaborative Group on Hormonal Factors in Breast Cancer, 2019). Infertility in FHA women is presently treated with either monofollicular ovarian stimulation followed by timed intercourse or intrauterine inseminations or by IVF/ICSI and, while highly effective, the latter is costly, time-consuming and burdensome (Katz et al., 2011).

To improve counselling and provide guidance for future research, this review summarises available knowledge on the factors allowing the return of menses. While focussed primarily on clinical data, we also include a review of the mechanistic and translational aspects of anovulation and the return of menses gained from animal studies. We evaluate (i) which factors determine the recovery of menses and its timing, (ii) how these factors exert their physiological effects and (iii) whether there are any useful therapeutic options to induce recovery.

Methods

A systematic review of epidemiological data on resumption of menses in humans was performed in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) criteria (Liberati et al., 2009). A PubMed search from 1966 to February 2020 included the MeSH terms: ‘re-onset/return or recovery of menses/menstruation’, ‘anovulation’, ‘amenorrhoea’, combined with ‘hypothalamus’, ‘pituitary’, ‘gonadotropin releasing hormone (GnRH)’, ‘GnRH pulse generator’, ‘leptin’, ‘ghrelin’, ‘glucose’, ‘insulin’, ‘cortisol’, ‘kisspeptin’ AND/OR ‘influencing factors’, ‘body weight’, ‘energy balance’, ‘eating disorders’, ‘nutrition’, ‘diet’, ‘sports’, ‘stress’, ‘mental health’, ‘posttraumatic stress disorder (PTSD)’, ‘genetics’ and/or ‘treatment’, ‘medication’, ‘stress management’, ‘psychotherapy’ and ‘cognitive behavioural therapy (CBT)’. A ‘return of menses’ set element AND/OR a ‘factor/physiology’ set element were always included in title, abstract or all fields, where capital letters indicate Boolean connectors. All studies were screened by title and abstract. Eligible articles in English, German and French were read, and the relevant information was extracted. Secondary literature from these studies was evaluated for further relevant publications.

FHA is considered to be chronic anovulation not due to any identifiable anatomic or organic causes, i.e. it is a diagnosis of exclusion without any generally accepted diagnostic criteria (Bomba et al., 2007; Genazzani et al., 2010; Gordon, 2010; Sowinska-Przepiera et al., 2015; Gordon et al., 2017). The differential diagnosis between amenorrhoea and FHA has recently been reviewed (Gordon et al., 2017). Polycystic ovary syndrome (PCOS) was considered a diagnosable cause of amenorrhoea and therefore not included within the present review.

Any study reporting at least one menstruation after a phase of FHA was considered within this review. The initial systematic search yielded a total of 1424 manuscripts published between 1966 and February 2020, with 49 manuscripts directly addressing either recovery of menses (N = 34) or prolonged amenorrhoea (N = 15) in humans. Prospective (N = 23) and cross-sectional (N = 13) studies investigating between 12 and 463 participants formed the majority of research. In addition, five retrospective studies, six case-control studies and two case reports were analysed.

Quality of observational studies was assessed using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria (von Elm et al., 2007). Case reports were evaluated by the Consensus-based Clinical Case Reporting Guideline Development (CARE) Guidelines (Gagnier et al., 2013).

Animal studies investigating the neural mechanisms underlying the impact of metabolic factors on the reproductive axis were considered by author A.E.H. To ensure, no major paper was omitted, recent reviews on the subject from three different research groups were consulted (Evans and Anderson, 2017; Chianese et al., 2018; Ronnekleiv et al., 2019; Vazquez et al., 2019).

Methodological details of included studies

Resumption of menses and/or risk factors for prolonged amenorrhoea were investigated predominantly after the treatment of eating disorders (N = 29), but also in athletes (N = 13) and in stress-related FHA (N = 7). Most studies compared anthropometric data (body weight, BMI, body fat) or hormonal parameters between women with and without return of menses aged <30 years. Out of the 34 studies directly investigating the recovery of menses, 12 controlled for well-known hormonal causes of cycle disturbances and related diseases such as PCOS, thyroid diseases, hyperprolactinaemia, adrenogenital syndrome, disturbances of the adrenal and pituitary glands and premature ovarian insufficiency. The underlying pathophysiology and the timeline of recovery of menses were rarely addressed. Of the 47 observational studies, 13 were low, 29 were medium and 5 were high quality (von Elm et al., 2007). The quality of case reports was middle (Kopp-Woodroffe et al., 1999) to high (Mallinson et al., 2013).

Results

We first define the diagnostic criteria for FHA and for the resumption of menses before summarising the causes of FHA, as their reversal may play a role in the return of menses. We next address potential physical and psychological mechanisms involved in the onset and end of FHA including their interactions. Finally, we will give an overview of currently available treatment options.

Diagnosis of functional hypothalamic amenorrhea (FHA)

To diagnose FHA, 16 studies relied solely on self-reported amenorrhoea (Table I) while 30 studies evaluated initial cycle status by serum (oestradiol, FSH and LH) or urine (oestrone glucuronide, pregnanediol glucuronide and LH) hormonal parameters. Transvaginal ultrasound of the ovaries was performed in four studies and radiography of the sella turcica was performed in one (Falsetti et al., 2002). A positive response to a GnRH stimulation test was another diagnostic criterion (Pentz and Nakic Rados, 2017). FHA was diagnosed after at least three (16 studies) or six (8 studies) months of amenorrhoea. Further prerequisites for the diagnosis of FHA were regular menses prior to FHA (Christo et al., 2008; Sterling et al., 2009) or a coincidence of amenorrhoea with increased exercise or low weight (Welt et al., 2004). Four studies provided no specific definition of FHA (Holtkamp et al., 2003; Bodell and Mayer, 2011; Chou et al., 2011; Rigaud et al., 2011).

Table I

Diagnostic criteria for FHA and return of menses.

Author, year	Diagnosis of FHA							Diagnosis of return of menses				Quality2
	Self- reported	Hormones		US	Amenor- rhea	Exclusion criteria1	No definition	Self- reported	Nb cycles/ cycle length	Control of ovulation	No definition
		Serum Urine
Abbate Daga et al., 2012	X	X	–	–	–	X	X	–	–	–	X	Middle
Arends et al., 2012	X	X	–	–	3 months	E, P	–	X	3/ < 36 days	–	–	Middle
Arimura et al., 2010	X	X	–	–	–	–	X	X	–	–	X	Middle
Audi et al., 1998	X	X	–	–	–	–	X	X	–	–	X	Low
Berner et al., 2017	X	–	–	–	3 menses	–	–	X	–	–	X	Middle
Bodell and Mayer, 2011	–	–	–	–	–	–	X	X	–	–	X	Middle
Brambilla et al., 2003	X	X	–	X	–	X	X	–	–	–	X	Middle
(,Chou and Perry, 2013)	–	X	X	–	–	–	X	X	Not >37.5% of normal length	X	–	High
Christo et al., 2008	X	X	–	–	3 menses	X	–	–	–	–	X	Middle
Cialdella-Kam et al., 2014	X	X	–	–	–	–	–	X	–	X	–	Middle
Cominato et al., 2014	X	X	–	–	–	X	X	X	–	–	–	Middle
Dei et al., 2008	X	X	–	–	–	–	X	X	–	–	X	Middle
Dempfle et al., 2013	X	–	–	–	–	X	X	X	–	–	X	Middle
El Ghoch et al., 2016	X	–	–	–	3 menses	X	–	X	3 menses last 6 months	–	–	Middle
Falsetti et al., 2002	X	X	–	X	6 months	X	–	X	–	X	–	Middle
Faust et al., 2013	X	–	–	–	–	–	X	X	1	–	–	Middle
Favaro and Santonastaso, 2009	X	–	–	–	–	–	X	–	–	–	X	Low
Genazzani et al., 1995	X	X	–	–	6 months	X	–	X	–	–	X	Low
Genazzani et al., 2012	X	X	–	–	6 months	X	–	X	–	–	X	Low
Giles and Berga, 1993	X	X	–	–	–	X	X	–	–	–	X	Middle
Golden et al., 1997	X	X	–	–	3 months	X	–	X	2	–	–	Middle
Golden et al., 2008	X	–	–	–	3 months	X	–	X	2	–	–	Low
Holtkamp et al., 2003	–	–	–	–	–	–	X	X	–	–	X	Low
Jacoangeli et al., 2006	X	X	–	–	–	–	X	X	–	–	X	Low
Johnson and Whitaker, 1992	X	–	–	–	3 months	X	–	–	–	–	X	Low
Karountzos et al., 2017	X	X	–	–	–	X	–	X	2	–	–	High
Kohmura et al., 1986	X	–	–	–	–	–	X	X	–	–	X	Low
Kopp-Woodroffe et al., 1999	X	–	–	–	3 months/ 3–4 menses	E	–	X	–	X	–	Middle
Lagowska et al., 2014	X	X	–	–	–	X	X	X	–	–	–	Middle
Laughlin and Yen, 1996	X	X	–	–	6 months	X	–	–	–	–	–	Middle
Mallinson et al., 2013	X	X	X	–	3 months	X	–	X	2/ < 36 days	X	–	High
Marcus et al., 2001	X	–	–	–	–	X	X	–	–	–	X	Middle
Martini et al., 2016	X	–	–	–	3 months	Other ED* than AN	–	–	–	–	X	Low
Melin et al., 2016	X	–	–	–	3 menses	X	–	–	–	–	–	Middle
Miller et al., 1998	X	X	–	–	3 months	X	–	–	–	–	–	Middle
Miller et al., 2004	X	X	–	–	3 months	E	–	–	–	–	–	Middle
Misra et al., 2006	X	X	–	–	–	< 15.3 y	X	X	3 in last 6 months	–	–	High
Pentz and Nakic Rados, 2017	X	X	–	–	6 months	X	–	–	–	–	–	Middle
Peric et al., 2016	X	–	–	–	–	–	–	–	–	–	–	Low
Pitts et al., 2014	X	X	–	–	–	–	–	X	–	–	–	Low
Reed et al., 2015	X	X	X	–	3 months/ <6 menses per year	X	–	X	≥ 1	–	–	Middle
Rigaud et al., 2011	–	–	–	–	–	–	X	X	–	–	X	Middle
Shen et al., 2013	X	X	–	–	6 months	X	–	X	3/ 21–35 days	–	–	Low
Sterling et al., 2009	X	–	–	–	3 months	X	–	X	3	–	–	Middle
Swenne et al., 2004	X	–	–	–	3 months	E	–	X	–	–	X	Middle
Tinahones et al., 2005	X	X	–	X	–	X	X	X	–	–	X	Middle
Latzer et al., 2019	X	X	–	–	–	X	X	X	2	–	–	High
Welt et al., 2004	X	X	X	X	6 months	X	–	X	–	X	X	High
Winkler et al., 2017	X	–	–	–	6 months	E	–	X	–	–	X	Middle

1

Exclusion criteria involved known hormonal causes of cycle disturbances (elevated levels of androgens, cortisol, or prolactin) and diseases (i.e. polycystic ovary syndrome (PCOS), thyroid diseases, subfunction of the adrenal gland, pituitary gland defects). Studies with only one or two simple exclusion criteria were described separately.

2

Quality of observational studies was assessed using STROBE criteria (von Elm et al., 2007). Case reports were evaluated by the CARE Guidelines (Gagnier et al., 2013) *Abbreviations: AN (anorexia nervosa), E (estrogen intake), ED (eating disorder), GnRH (gonadotropin-releasing-hormone), LH (luteinising hormone), P (pregnancy), y (year), FHA (functional hypothalamic amenorrhea), US (Ultrasound).

Diagnostic criteria for resumption of menses

The majority of studies (N = 24) did not provide any definition of resumption of menses (Table I). Of the 34 studies explicitly addressing resumption of menses, 31 relied on self-reported bleeding with 12 differentiating between one, two or three consecutive spontaneous menstrual cycles as cut-off to diagnose recovery of menses. Four studies required a cycle length of <36 days to diagnose recurrence of menses. Further criteria were the exclusion of other potential causes of vaginal bleeding such as infection or trauma. Six studies evaluated either with urine or serum hormonal parameters or by ultrasound whether menstrual cycles were ovulatory.

Body weight, energy availability and return of menses

The association between body weight, energy availability and return of menses was explored in eating disorders, or in the context of physical activity (Tables II and III, Fig. 1). The majority of the 19 studies support that anthropometric characteristics, i.e. BMI, body weight or body composition, are associated with resumption of menses (Table II). However, there are also conflicting results (Arimura et al., 2010; El Ghoch et al., 2016).

Figure 1

Timing and prevalence of resumption of menses after different causes of prolonged amenorrhoea. Data in women with eating disorders are presented in blue; data in the context of physical activity are presented in red. Data from case series of two * and three ** women.

Table II

Criteria allowing resumption of menses.

Study	N (FHA)	Cause FHA1	BMI kg/m2 at mens (Mean ± SD)	BMI increase (kg/m2) (Mean ± SD)	Body weight (kg) (Mean ± SD)	Weight gain (kg)	Median body weight (Mean ± SD )	Body fat at presence of menstrual cycle (%, unless otherwise specified)	Other factors for resumption of menses	Quality2
Arends et al., 2012	42	PE	22.7 ± 0.6	1.9 ± 0.4	58.0 ± 2.7	5.3 ± 1.1	–	–	Energy intake	Middle
Arimura et al., 2010	20	ED	17.8 ± 0.9	1.8 ± 1.3	43.7 ± 4.1	11.8 ± 3.7	85.1 ± 4.1	24.4 ± 4.2	–	Middle
Berner et al., 2017	69	ED	Not reported	–	–	–	–	–	BMI at onset of amenorrhoea	Middle
Cialdella-Kam et al., 2014	7	PE	22.9 ± 2.5	–	64.0 ± 8.0	1.6	–	24.1 ± 3.9	Amenorrhoea < 8 months et al., improved energy status	Middle
Dei et al., 2008	43	ED	19.2 ± 1.8	0.1 ± 1.4	52 (48 - 55)	7 (3.5–11)	–	11.7 ± 3.5 (kg)	–	Middle
Dempfle et al., 2013	80	ED	18.8 ± 1.7	–	–	–	91.1 ± 8.7	–	15th - 20th BMI percentile3	High
El Ghoch et al., 2016	35	ED	19.9 ± 0.8	–	–	–	–	27.7 ± 5.7	–	Middle
Faust et al., 2013	84	ED	Not reported	–	–	6.7 ± 4.9	94.9 ± 9.3	–	Higher scores in Eating Disorder Examination – Questionnaire4	Middle
Favaro and Santonastaso, 2009	248	ED	18.7 ± 1.7	–	–	–	–	–	Level of hostility et al., season	Low
Golden et al., 1997	100	ED	19.2 ± 1.8	–	50.4 ± 5.2	> 2.05 kg than at beginning of amenorrhea	90.5 ± 8.5	20.6 ± 3.6	–	Middle
Golden et al., 2008	56	ED	19.0 ± 1.6	–	49.5 ± 4.7	–	90.4 ± 8.4	–	Mean BMI percentile3: 27.1	Low
Jacoangeli et al., 2006	10	ED	19.1 ± 2.7	–	–	–	–	24.4 ± 11.2	Less Exercise: 14 ± 20 vs. 45 ± 38 (min/d)	Low
Karountzos et al., 2017	25	ED	19.6 ± 0.6	–	–	7.5 ± 1.5	–	22.6 ± 2.4	Total body/trunk fat > 20%	High
Kohmura et al., 1986	35 (21)	ED	–	–	–	–	90.2 ± 4.3	–	–	Low
Kopp-Woodroffe et al., 1999	4	PE	20.6 (Median)	–	–	–	–	18.4 (Median)	Energy balance	Middle
Lagowska et al., 2014	45 (31)	PE	20.7 ± 1.5	–	59.6 ± 5.3	–	–	21.0 ± 3.5	Energy balance	Middle
Mallinson et al., 2013	2	PE	22.0/20.7	–	54.7/54.0	2.5/1	–	20/20.24	Energy balance	High
Misra et al., 2006	33	ED	20.0 ± 2.1	3.2 ± 2.2	–	–	–	25.5 ± 4.8	–	High
Pitts et al., 2014	37	ED	Not reported	2.2 (Mean)	–	–	–	23.1 (Mean)	Body fat at baseline: 18.0 ± 5.0	Low
Reed et al., 2015	30	PE	21.7 ± 0.2	–	58.8 ± 0.8	–	–	25.7 ± 0.6	Energy balance	Middle
Rigaud et al., 2011	463	ED	>18.5 (87%)*	–	–	–	–	–	Higher fat and energy intake/Decrease in physical hyperactivity	Middle
Swenne et al., 2004	127	ED	19.2 ± 1.7	0.6 ± 0.9	–	–	–	–	–	Middle
Tinahones et al., 2005	40	ED	19.6 ± 2.6	–	51.7 ± 7.4	–	–	21.8 ± 9.4	–	Middle
Tokatly et al., 2019	20	ED	20.2 ± 1.1	–	52.0 ± 5.6	–	–	24.6 ± 3.5	–	High
Winkler et al., 2017	52	ED	21.2 ± 2.8	5.9 ± 3.2	–	–	–	26.8 ± 6.3	–	Middle

1

Cause of functional hypothalamic amenorrhoea (FHA): physical exercise (PE), eating disorder (ED).

2

Quality of observational studies was assessed using STROBE criteria (von Elm et al., 2007). Case reports were evaluated by the CARE Guidelines (Gagnier et al., 2013).

3

BMI percentiles derived from 17 German studies (Dempfle et al., 2013) or different national health surveys (Kuczmarski et al., 2000).

4

Global score and in the categories concern/restrain. The EDE-Q 6.0 is a 28–item measure (© 2008 by Christopher G. Fairburn and Sarah Beglin) derived from the EDE (Fairburn et al., 2008). The EDE-Q is scored using a 7-point, forced-choice rating scale (0–6) with scores of 4 or higher indicative of clinical range. The subscale and global scores reflect the severity of eating disorder psychopathology. A global score is the sum of the four subscale scores divided by the number of subscales (i.e. four).

*

In 87% of the women who reached a BMI > 18.5 kg/m2, menses resumed.

Table III

Timing of recovery.

	Author	N (FHA)	Type of study	Duration FHA (months) (Mean ± SD)	Intervention	Time of evaluation till return of menses	Percentage of women with return of menses	Time until return of menses (months) (Mean ± SD)	Quality1
Eating disorder	Abbate Daga et al., 2012	184	Prospective cohort	Not reported	Weight recovery, psychoeducational treatment	12 months	38.0% (AN*), 52.0% (EDNOS*), 54.0% (recovered AN), 74.0% (FHA without ED*)	Not reported	Middle
	Arimura et al., 2010	20	Case-control	Not reported	Weight recovery, cognitive behavioural therapy	Not reported	45.0%	1.6 ± 1.1 after reaching 85% of standard body weight2	Middle
	Berner et al., 2017	69	Cross-sectional	Not reported	Not reported	Not reported	26.0%	Not reported	Middle
	Bodell and Mayer, 2011	22**	Prospective cohort	Not reported	Weight recovery	12 months	36.0%	Not reported	Middle
	Dei et al., 2008	43	Cross-sectional	Not reported	Weight recovery, psychotherapy	Not reported	44.2%	Not reported	Middle
	Dempfle et al., 2013	80	Prospective cohort	Not reported	Weight recovery, cognitive-behavioural individual + group therapy, individual family sessions, group psycho-education for parents	12 months	47.0%	Not reported	High
	El Ghoch et al., 2016	35	Prospective cohort	Not reported	Weight recovery, cognitive behavioural therapy	12 months	35.0%	Not reported	Middle
	Faust et al., 2013	84	Retrospective	Not reported	Weight recovery, family-based treatment	12 months	67.9%	3.1 ± 2.5	Middle
	Favaro and Santonastaso, 2009	248	Prospective cohort	Not reported	Not reported	6–65 months	32.0%	Not reported	Low
	Golden et al., 1997	100	Prospective cohort	11.3 ± 11.6	Weight recovery, psychiatric intervention	2 years	Year 1: 68.0%/ Year 2: 95.0%	9.4 ± 8.2	Middle
	Golden et al., 2008	56	Retrospective	>3	Weight recovery, psychiatric intervention	12 months	64.3%	Not reported	Low
	Jacoangeli et al., 2006	250/10	Cross-sectional/Retrospective	44 ± 3 (A)/ 23 ± 23.7 (M)	Weight recovery	6 months	86.0%	Not reported	Low
	Karountzos et al., 2017	25	Prospective cohort	21.2 ± 2.7 (A)/16.7 ± 2.5 (M)	Weight recovery	8–26 months	58.0%	13.3 ± 2.9 (8–24)	High
	Kohmura et al., 1986	35 (21)	Prospective cohort	21.0 ± 3.7	Weight recovery, supportive psychotherapy	10 years (8–13)	81.0%	Not reported	Low
	Misra et al., 2006	33	Prospective cohorts	10.6 ± 10.2	Weight recovery	12 months	65.5%	Not reported	High
	Pitts et al., 2014	37	Prospective cohort	25 ± 23 (A)/ 18 ± 29 (M)	Not reported	18 months	76.0%	3–18	Low
	Rigaud et al., 2011	463	Prospective cohort	Not reported	Weight recovery, psychotherapy, cognitive behavioural therapy	13 years	97.0%	Not reported	Middle
	Swenne et al., 2004	127	Retrospective	Not reported	Weight recovery	12 years	74.4%	14 ± 12	Middle
	Tinahones et al., 2005	40	Prospective cohort	Not reported	Psychotherapy, no intense physical exercise	Not reported	100%	7.8 ± 7.3	Middle
	Tokatly Latzer et al., 2019	20	Prospective cohort	11.8 ± 8.9 (A)/8.7 ± 9.2 (M)	Weight recovery, psychiatric and psychological intervention	6 years	68.0%	1.9 ± 0.9 after reaching the target weight3	High
	Winkler et al., 2017	52	Cross-sectional	Not reported	Not reported	Not reported	54.0%	Not reported	Middle
Sports	Arends et al., 2012	42	Retrospective	Not reported	None	5 years	23.1%	17.7 ± 4.8	Middle
	Cialdella-Kam et al., 2014	7	Prospective cohort	<1 year in 5 and >1 year in 2 women	Nutrition supplement4	6 months	100%	2.6 ± 2.2	Middle
	Kopp-Woodroffe et al., 1999	4	Case reports (N = 4)	6, 3, 6, 9	Nutrition supplement4, one day rest/week	20 weeks	75.0%	2.1, 3, 5.5	Middle
	Lagowska et al., 2014	45 (31)	Prospective cohort	>6 years	Nutrition supplement4	12 weeks	0%	Not applicable	Middle
	Mallinson et al., 2013	2	Case reports (N = 2)	3, 11	Nutrition supplement4	12 months	100%	0.8 (3 months of FHA), 2.4 (11 months of FHA)	High
	Reed et al., 2015	30	Cross-sectional	>3 or ≤6 menses/year	None	Not reported	45.0%	Not reported	Middle

1

Quality of observational studies was assessed using STROBE criteria (von Elm et al., 2007). Case reports were evaluated by the CARE Guidelines (Gagnier et al., 2013).

2

Standard Body weight (SBW): Median body weight adjusted for age, gender, frame size and height (Frisancho 1984).

3

Target weight was estimated individually, per case, according to BMI percentiles and optimal individual height (Modan-Moses et al., 2012).

4

Carbohydrate-protein (CHO-PRO) supplement (360 kcal/day, 54 g CHO/day, 20 g PRO/day.

*

AN, anorexia nervosa; BN, bulimia nervosa; EDNOS, eating disorder not otherwise specified.

**

Menstrual status was not reported at baseline.

Eating disorders, body weight and energy availability

Patients who resume menses after eating disorders have a higher BMI (19 kg/m² vs. 17.5 kg/m²) or body weight than those who do not (Le Grange et al., 2012; Dempfle et al., 2013). As with individual differences at the onset of amenorrhoea, the individual BMI for recovery of menses differs strongly and comparison of different studies is hampered by the use of different weight measures. For example, two-thirds of anorectic girls became amenorrhoeic at a BMI between 17 and 18.9 kg/m² while the remaining third had a normal BMI (Berner et al., 2017). A higher BMI at the onset of amenorrhoea is associated with a higher BMI needed to allow return of menses (Pitts et al., 2014; Berner et al., 2017). At or above a BMI of 19 kg/m² and ≥23% body fat, about 50% of women are expected to resume menses (Tinahones et al., 2005), although women with a BMI of 14 kg/m² and 11% body fat still had a probability of 25% for recovery of menses (Winkler et al., 2017). Generally, recovery of menses seems to require about 2 kg more than the weight of the woman at the time at which menses were lost (Golden et al., 1997).

Using other weight parameters, menstrual recovery was reported to occur at 91.6 ± 9.1% of standard body weight (median body weight adjusted for age, gender, frame size and height) (Frisancho, 1984; Golden et al., 1997) or 94.9 ± 9.3% of expected body weight (EBW = optimal weight related to height and/or age for healthy nutritional status with the lowest rate of mortality) (Faust et al., 2013). In anorectic post-menarcheal girls, a BMI ≥27th percentile, based on percentiles derived from different national health surveys, or a BMI at the 24th percentile, based on BMI measurements from 17 German studies, seems to be necessary for recovery of menses (Kuczmarski et al., 2000; Golden et al., 2008; Dempfle et al., 2013). Interestingly, during spring or summer, anorectic women needed on average 2 kg less for resumption of menses than during fall or winter (Favaro and Santonastaso, 2009).

According to prospective data from anorectic patients, the onset and regularity of menstrual cycles require between 18% and 28% of body fat (Frisch and McArthur, 1974; Frisch, 1987; Misra et al., 2006; Pitts et al., 2014; El Ghoch et al., 2016; Karountzos et al., 2017; Winkler et al., 2017; Tokatly Latzer et al., 2019) (Table II). However, even with 36% body fat, not all women resumed menses (Tinahones et al., 2005). An increase of about 1% in total body fat at discharge after anorexia treatment was reported to augment the probability of menses by ∼14% per year (El Ghoch et al., 2016). Modifications in fat distribution following weight recovery seem to be irrelevant for resumption of menses (Dei et al., 2008; Mayer et al., 2009).

Independent from absolute weight, dieting is considered to be a risk factor for amenorrhoea (Martini et al., 2016). After at least 1 year of stable normal weight, caloric intake still differed significantly between women with persisting FHA and age-, weight-, as well as body fat-matched control women with a regular cycle (Miller et al., 1998). In addition, eating disorders are closely associated with stress (Keski-Rahkonen and Mustelin, 2016; Smith et al., 2018), which may further increase the risk for prolonged amenorrhoea (see below).

A few studies have reported no significant association between body fat and the return of menses (Golden et al., 1997; Jacoangeli et al., 2006; Dei et al., 2008; Arimura et al., 2010). Arimura et al. (2010) reported no resumption of menses immediately after weight recovery but this may have been too short a period of observation as studies supporting an association are based on a longer follow-up period (Misra et al., 2006; Pitts et al., 2014; El Ghoch et al., 2016; Karountzos et al., 2017; Tokatly Latzer et al., 2019). Golden et al. (1997) applied skinfold thickness measurements, which are of rather limited diagnostic quality to evaluate body fat (El Ghoch et al., 2012). The cross-sectional studies reporting no association were either small (Jacoangeli et al., 2006) or applied an arbitrary definition of weight recovery (Dei et al., 2008).

In summary, the absolute BMI has to rise to between 17.7 ± 1.4 kg/m² (Pitts et al., 2014) and 22.9 ± 2.5 kg/m² (Cialdella-Kam et al., 2014) and body fat has to rise to between 18% and 28% to allow resumption of menses. The BMI at occurrence of amenorrhoea is important in setting the individual weight target for recovery of menses. As current research on the association between eating disorders has focused on weight, information is lacking on whether specific eating disorder-related malnutrition is related to amenorrhoea.

Prognosis and timing of return of menses after weight recovery. Between 35% and 54% of women experience return of menses immediately after achieving normal weight (Arimura et al., 2010; Bodell and Mayer, 2011; Dempfle et al., 2013; El Ghoch et al., 2016; Winkler et al., 2017). After a longer treatment/observation period, these rates may rise to 80% or even 100% (Kohmura et al., 1986; Golden et al., 1997; Tinahones et al., 2005; Jacoangeli et al., 2006) (Table III). Studies providing exact information on the time to recovery of menses after weight restoration reported a broad range from 50 ± 33 days up to 14 ± 12 months until the return of menses (Golden et al., 1997; Swenne, 2004; Tinahones et al., 2005; Arimura et al., 2010; Faust et al., 2013; Karountzos et al., 2017; Tokatly Latzer et al., 2019).

Between 5% and 68% of women have been reported to remain amenorrhoeic after weight recovery (Kohmura et al., 1986; Golden et al., 1997; Jacoangeli et al., 2006; Misra et al., 2006; Favaro and Santonastaso, 2009; Dempfle et al., 2013; Pitts et al., 2014; El Ghoch et al., 2016; Karountzos et al., 2017; Tokatly Latzer et al., 2019) with between 5% and 14% ultimately remaining amenorrhoeic (Golden et al., 1997; Jacoangeli et al., 2006).

According to the guidelines of the Endocrine Society, ‘the term “functional” hypothalamic amenorrhea implies that correction or amelioration of the causal factors will restore ovulatory ovarian function’ (Gordon et al., 2017). However, the return of menses does not necessarily occur after such correction. In previously anorectic adolescents, 86% experienced the return of menses within 6 months of stable weight (Jacoangeli et al., 2006). After 1 year of successful treatment against anorexia, 35–68% of women resumed menses and 95% had recovered by year 2 (Golden et al., 1997; Misra et al., 2006; Golden et al., 2008; Bodell and Mayer, 2011; Abbate Daga et al., 2012; Dempfle et al., 2013; Faust et al., 2013; El Ghoch et al., 2016). As all adolescents with premenarchal onset of anorexia remained amenorrhoeic at the 12-month follow up following weight restoration, the premenarchal onset of eating disorders seems to be an unfavourable factor for recovery of menses (Dempfle et al., 2013). About 5% of previously anorectic adolescents remain amenorrhoeic 2 years after achieving normal weight (Golden et al., 1997). In the study with the longest observation period of 13 years, 97% of anorectic women had resumed their menses (Rigaud et al., 2011). In that study, a BMI >18.5 kg/m² and lack of physical hyperactivity explained 67% of the variance in return of menses.

Although return of menses clearly depends on prerequisites such as BMI or energy availability, individual recovery is difficult to predict. None of the available mathematical models succeeds in predicting the exact time for return of menses. Per cent body fat and BMI have equal predictive quality but explain only 14% of the variation in recovery of menses (Winkler et al., 2017).

Excessive exercise, body weight and energy availability

Total absolute and percentage weight gain, as well as BMI, are found to differ between athletes with and without recovery of menses (Arends et al., 2012). Nutrition and energy balance also play a role in the return of menses; amenorrhoeic athletes eat less fat but more carbohydrates and fibre than those with a normal menstrual cycle (Laughlin et al., 1998; Cialdella-Kam et al., 2014). Energy availability (defined as the net input of energy remaining after exercise training and energy needed for all other metabolic processes, normalised to kilogram of lean body mass (Loucks and Thuma, 2003) is lower in amenorrhoeic than in eumenorrhoeic exercising women (Williams et al., 2001), but does not differentiate between ovulatory and anovulatory cycles (Reed et al., 2015). Athletes with amenorrhoea often consume too few calories for their energy needs (Melin et al., 2016; Elliott-Sale et al., 2018) and the likelihood of exercise-related menstrual abnormalities seems to vary with the magnitude of the energy deficit (Williams et al., 2015). LH pulsatility has been reported to be disrupted when energy intake in women is <30 kcal/kg lean body weight per day (Dueck et al., 1996; Loucks and Thuma, 2003; Loucks et al., 2011; Reed et al., 2015). Also, an overall reduction of energy by 470 and 810 kcal per day in women with an initial body fat between 15% and 35% and a BMI 18–25 kg/m² increases the risk for menstrual cycle disturbances (Williams et al., 2015). Menstrual function may already stop at energy availabilities above this threshold, but specific causes for these differences have not be identified as yet (Reed et al., 2015; Lieberman et al., 2018; Holtzman and Ackerman, 2019). It is possible that age-related physiological differences during adolescence may influence actual parameter thresholds. However, very few studies have so far addressed the identification of clear cut-off values and related influencing factors. Unfortunately, the exact extent of physical activity and details of nutrition and nutritional supplements are only assessed in a few studies, which makes comparison of results difficult.

Athletes with FHA also show a significantly lower resting energy expenditure compared to eumenorrhoeic athletes, however, part of this difference may result from discrepancies in initial and actual body weight or BMI between study participants (Christo et al., 2008; Sterling et al., 2009).

Prognosis and timing of the return of menses after adjusting energy and weight. Weight gain or increased energy availability by nutritional supplements, and a decrease in energy expenditure, increases the chances for resumption of menses (Table III) (Kopp-Woodroffe et al., 1999; Arends et al., 2012; Mallinson et al., 2013; Cialdella-Kam et al., 2014; Lagowska et al., 2014). In athletes close to normal weight, amenorrhoea may reverse when training is reduced (Warren, 1980; Benson et al., 1996). With adequate calorie-intake, menses are expected to reoccur in 75–100% of women (Kopp-Woodroffe et al., 1999; Mallinson et al., 2013; Cialdella-Kam et al., 2014). An appropriately balanced diet with the simultaneous limitation of training volume and intensity is therefore the main tool to reduce menstrual disorders in athletes (Manore et al., 2007; Nattiv et al., 2007) with per cent weight gain being a significant positive predictor for recovery of menses (Cialdella-Kam et al., 2014).

After adjusting weight and energy balance, the time until recovery of menses varied from 11 weeks to 33 months (Kopp-Woodroffe et al., 1999; Arends et al., 2012; Mallinson et al., 2013; Cialdella-Kam et al., 2014). While one study showed a correlation between the duration of amenorrhoea and recovery of menses (Cialdella-Kam et al., 2014), another study did not support such association (Arends et al., 2012).

Comparison of women with eating disorders to women with excessive exercise

The relevance of different factors involved in the resumption of menses seems to vary in relation to factors involved in the initiation of amenorrhoea. In eating disorders, the extent of weight and/or fat gain seems to be particularly important for recovery (Dempfle et al., 2013; El Ghoch et al., 2016) whereas in normal-weight athletes, adequate energy intake seems to play the major role (Reed et al., 2015). Women with eating disorders show addictive and obsessive-compulsive traits that can manifest in excessive physical activity (Davis and Claridge, 1998). Amenorrhoea is more prevalent among athletes with eating disorders (Peric et al., 2016), and 39–48% of women with eating disorders also engage in excessive exercise, i.e. there is a common overlap of risk factors for prolonged amenorrhoea (Freimuth et al., 2011).

Potential mechanisms underlying the return of menses

Body weight and energy availability

There is a well-documented regulatory influence of energy balance on fertility. For the most part, this is brought about by circulating energy-related hormones and metabolites that modulate the functioning of central hypothalamic networks controlling the secretion of GnRH.

Mammalian fertility is governed by a neural network that integrates a range of internal and external cues to control the release of GnRH that, in turn, generates pulsatile and surge profiles of gonadotropin secretion (Fig. 2) (Herbison, 2016). Although pulsatile gonadotropin secretion occurs throughout the menstrual cycle, the frequency and amplitude of pulses change across the cycle to ensure the correct maturation of developing follicles (Herbison, 2018). Studies in animal models have now demonstrated that a population of kisspeptin neurons located in the hypothalamic arcuate/infundibular nucleus operate as the ‘GnRH pulse generator’ by activating the GnRH neurons to generate pulsatile gonadotropin secretion (Herbison, 2018; Plant, 2019). It is clear from both animal and human studies that chronic and acute energetic stressors can result in a marked reduction in LH pulsatility (Loucks et al., 1998; Hilton and Loucks, 2000; Loucks and Thuma, 2003; Wade and Jones, 2004). In addition to pulses, the hypothalamus and pituitary generate the mid-cycle LH surge that initiates ovulation. Disturbances of either pulsatile or surge profiles of gonadotropin hormone secretion can suppress fertility.

Figure 2

Schematic representation of the regulation of the GnRH pulse and surge generators in mammalian species. Factors that regulate GnRH neuron function across species and including humans are represented in direct connection with the modality. For example, metabolic factors and stress suppress the GnRH pulse generator in all species. In contrast, species-specific factors are shown as connecting with the appropriate modality by arrows. Reprinted with permission from Herbison (2016).

Investigators have now discovered a wide range of circulating factors that inform the brain on the metabolic and energy status of the body. Whilst these hormones are primarily driving appropriate central energy regulation through appetite and energy expenditure, the same signals are thought to be used to regulate the GnRH neuron network and, accordingly, the menstrual cycle (Fernandez-Fernandez et al., 2006; Navarro and Kaiser, 2013; Evans and Anderson, 2017). Discussed below are the roles of leptin, ghrelin and insulin, considered to be the primary peripheral factors signalling information on body weight and energy availability to the GnRH neuron network. The impact of activating stress pathways, often not easily separated from the energy deficit itself, is also considered.

The vast majority of work in this field has examined how energy insufficiency/stress operates to suppress pulsatile gonadotropin secretion, resulting in cycle abnormalities and infertility. Unfortunately, very little work has examined the return of fertility with the presumption being that it is a reversal of the mechanism that initiates the infertility generated by energy stress. Perhaps surprisingly, the resumption of pulsatile LH secretion in re-fed, acutely energetically stressed women is much slower (taking up to 1 week) than in men or in experimental animals examined to date (Cameron, 1996; Loucks et al., 1998; Szymanski et al., 2007).

Leptin. Secreted by adipocytes into the circulation, leptin signals body fat stores with elevated levels acting on the hypothalamus to reduce feeding and increase energy expenditure. Leptin concentrations also fluctuate on a shorter time scale unrelated to weight or body fat, being reduced with fasting (Ahima et al., 1996; Grinspoon et al., 1997; Flier, 1998). Women with anorexia and exercise-induced amenorrhoea are hypoleptinemic (Mantzoros et al., 1997; Audi et al., 1998; Miller et al., 1998; Jimerson et al., 2000). Interestingly, even after controlling for body fat, women with exercise- or eating-related FHA have significantly lower leptin levels than those of their ovulatory counterparts (Miller et al., 1998; Warren et al., 1999; Andrico et al., 2002). A leptin level of 1.85 ng/ml (Reference 4.1—25 ng/ml) appears to be the critical level for amenorrhoea (Kopp et al., 1997). Values above this lead to an increase of LH (Holtkamp et al., 2003) and more than 20% of amenorrhoeic women recover menses at levels >1.85 ng/ml (Tinahones et al., 2005). However, a cross-sectional study found no differences in leptin levels or BMI between two groups of anorectic patients with and without amenorrhoea (Audi et al., 1998). Furthermore, the reinstatement of pulsatile LH secretion by refeeding chronically food-restricted ewes was found to be unrelated to circulating leptin concentrations (Nakamura et al., 2010).

The mechanisms through which leptin controls fertility are multifaceted and may involve peripheral as well as central actions. For example, high leptin levels enhance oocyte nuclear and cytoplasmic maturation and affect follicle rupture and corpus luteum formation (Ruiz-Cortes et al., 2003; Craig et al., 2004). However, studies in rodents indicate that the primary impact of leptin on fertility likely arises through central mechanisms. The deletion of leptin receptors from just the brain results in infertility (Quennell et al., 2009) and actively blocking leptin signalling in the brain reduces pulsatile LH secretion (Carro et al., 1997). The effects of leptin are often considered to be permissive in the sense that they enable normal functioning of the GnRH neuron network rather than actually determining its magnitude or mode of action.

Many studies in animal models have tried to establish the neural pathway through which leptin influences GnRH neurons. As GnRH and kisspeptin neurons seem unlikely to express functionally significant leptin receptors themselves, attention has focussed upon indirect mechanisms by which circulating leptin modulates the activity of neurons that project to and control the kisspeptin and/or GnRH neurons (Navarro and Kaiser, 2013; Evans and Anderson, 2017). While roles have been proposed for leptin to operate through premammilary nucleus neurons (Donato et al., 2011) and GABA neurons in the brain (Zuure et al., 2013) to control GnRH secretion, most evidence favours an effect of leptin on neuropeptide Y/agouti-related peptide (NPY/AgRP) neurons (Ronnekleiv et al., 2019). These are the same cells implicated in the potent actions of leptin on energy metabolism. As such, it is envisaged that low leptin levels act through NPY/AgRP neurons to both increase feeding and disable the normal menstrual/oestrous cycle. Precisely, how the leptin-sensing NPY/AgRP neurons impact upon GnRH secretion remains unclear with the most likely pathway being through the direct control of the kisspeptin neuron pulse generator (Hessler et al., 2020). The resumption of menses in FHA women may result from the re-establishment of normal energy balance leading to normalised leptin levels and consequently reduced NPY/AgRP neuron activity permitting normal pulsatile GnRH secretion.

Insulin and glucose. Diet and negative energy balance associated with FHA generate a hypometabolic state that includes, among other abnormalities, lowered circulating insulin levels (Laughlin and Yen, 1996). As such, insulin is one potential pathway through which the menstrual cyclicity is regulated in women with exercise- or diet-related FHA. Indeed, fasting insulin levels are higher in women with a history of eating disorders after return of menses compared to women remaining amenorrhoeic (Tinahones et al., 2005; Dei et al., 2008; Karountzos et al., 2017). A similar association has been confirmed following weight recovery (defined as an increase in weight >85% of the initial weight before amenorrhoea) (Dei et al., 2008), during the 8- to 26-month treatment phase for an eating disorder (Karountzos et al., 2017) and in the early follicular phase in women with resumption of menses (Tinahones et al., 2005). However, studies in male monkeys indicate that the key factor underlying the return of normal LH pulsatility following a nutritional stress is the increase in caloric intake independent of glucose or insulin (Cameron, 1996).

It is likely that insulin operates at multiple levels of the reproductive axis to exert a modulatory effect upon fertility. Unlike leptin, there is much less certainty that insulin actions in the brain are necessary or critical. Whereas an early study reported that the deletion of insulin receptors selectively from the brain of mice resulted in mild hypogonadism (Bruning et al., 2000), another study found no reproductive abnormalities (Evans and Anderson, 2017). Furthermore, the deletion of insulin receptors selectively from GnRH or kisspeptin neurons, or a range of other neuronal phenotypes, has been found to have no impact upon fertility in mice (Divall et al., 2010; Qiu et al., 2013; Evans and Anderson, 2017). Interestingly, over nutrition resulting in obesity-induced infertility in mice appears to be dependent, at least in part, upon insulin signalling at the GnRH neuron (Divall et al., 2010), pituitary gland (Brothers et al., 2010) and ovary (Wu et al., 2012). It is unknown whether this represents a similar mechanism to the cycle disturbances resulting from energy deficit.

Glucose concentrations themselves may represent an independent pathway through which energy stress suppresses the menstrual cycle. For example, reducing glucose availability within the brain suppresses pulsatile LH secretion in experimental animals (Murahashi et al., 1996; Lado-Abeal et al., 2002) and binge eating or binge-purge behaviour in humans is associated with a higher risk for FHA (Johnson and Whitaker, 1992). Differentiating any role of glucose from insulin in energy stress-evoked FHA can be challenging, but available evidence is equivocal regarding whether glucose may be more or less important than alterations in insulin secretion (Cameron, 1996; Szymanski et al., 2007; Roland and Moenter, 2011). Precisely, where glucose acts to modulate the menstrual cycle remains unclear given the ubiquitous requirement for glucose and multiple different sensors that could be involved. Peripherally, glucose deficiency can compromise the ability of the oocyte to reach the second metaphase, to extrude the first polar body (Dominko and First, 1997), or to achieve the blastocyte stage (Dan-Goor et al., 1997). There is also evidence demonstrating that glucose modulates the electrical excitability of GnRH neurons in a direct manner through AMPkinase, as well as indirectly through glucose-sensitive inputs from neurons located in the brainstem (Roland and Moenter, 2011). Thus, the return of a normoglycaemic state to women with FHA may contribute to the return of menses through direct glucose actions at multiple and varied sites throughout the reproductive axis.

Ghrelin. Ghrelin acts as a signal of starvation and energy insufficiency and is secreted by the stomach in a fluctuating pattern with elevated concentrations occurring prior to meals (Tena-Sempere, 2013). Unsurprisingly then, ghrelin concentrations are found to be continuously elevated in women suffering from chronic undernutrition and exercise-induced amenorrhoea (Tolle et al., 2003; De Souza et al., 2004; Christo et al., 2008; Schneider et al., 2008). In this case, an energy-deficient state results in elevated ghrelin levels (compared with reduced leptin, insulin and glucose concentrations) and evidence indicates that elevated ghrelin levels suppress gonadotropin secretion (Kluge et al., 2012). Hence, the elevated ghrelin levels found in women with FHA may represent another pathway contributing to their suppressed fertility (Christo et al., 2008).

In common with the other potential hormonal mediators highlighted above, ghrelin has multiple potential sites of action within the reproductive axis. Ghrelin receptors (GHS-R) are widely expressed in the ovary, pituitary and within the regions of the hypothalamus involved in the control of GnRH secretion (Gaytan et al., 2003, 2005; Tena-Sempere, 2007). However, animal studies indicate that ghrelin signalling is not itself essential for the suppression of cycles as the deletion of the ghrelin receptor in mice has no impact on fertility or feeding (Sun et al., 2003). It does, however, appear to modulate glucose sensing, insulin sensitivity and the stress response (Sun et al., 2008; Sominsky et al., 2017) and may, through these mechanisms, have some role in indirectly modulating the menstrual/oestrous cycle (Evans and Anderson, 2017).

Other hormones. While the discussion above considers what are thought to be the primary hormonal signals modulating fertility, women with energy-related FHA exhibit multiple other endocrine abnormalities. For example, free tri-iodothyronine (FT3), free thyroxine (FT4) and thyroid-stimulating hormone (TSH) are commonly decreased in amenorrhoeic women with eating disorders and these increase after resumption of menses (Tinahones et al., 2005; Dei et al., 2008; Karountzos et al., 2017). Whether and how thyroid hormones are related to the resumption of menses remains unclear. Similarly, levels of growth hormone (GH) and insulin-like growth factor-1 (IGF-1) are reduced in FHA (Miller et al., 2004; Bomba et al., 2007). An IGF-1 level of >342.8 ng/ml has been proposed to be a predictor for return of menses (Cominato et al., 2014), although two other studies have found no positive correlation (Falsetti et al., 2002; Arimura et al., 2010).

Psychological and psychogenic factors

The hypothalamic neurons regulating GnRH pulses not only respond to metabolic conditions but also to psychological and psychogenic factors (Drew, 1961; Fries et al., 1974; Frisch and McArthur, 1974; Mecklenburg et al., 1974; Bullen et al., 1985; Gadpaille et al., 1987; Berga and Girton, 1989; Loucks et al., 1989; Pirke et al., 1989; Warren et al., 1999; Roa et al., 2010; Berga and Naftolin, 2012; Garcia-Garcia, 2012; Sanchez-Garrido and Tena-Sempere, 2013; Castellano and Tena-Sempere, 2016).

Emotional stressors as measured by both subjective and objective parameters increase the risk for FHA in humans and in animals (Brown et al., 1983; Harlow and Matanoski, 1991; Facchinetti et al., 1993; Sanders and Bruce, 1999; Gordley et al., 2000; Kondoh et al., 2001; Bomba et al., 2007). In adolescent girls, changing school, initiating sexual activity, breaking up with a boyfriend, chronic illness, death of a friend or family member and family conflicts may result in FHA (Bomba et al., 2007). Psychological risk factors that chronically activate the HPA axis include perfectionism, high need for social approval, conditional love and/or unrealistic expectations of self and others (Berga and Girton, 1989; Giles and Berga, 1993; Marcus et al., 2001). College students who later became amenorrhoeic have been reported to be more anxious, stubborn and perfectionist (Shanan et al., 1965). It is possible that sleep deprivation may exacerbate this by further activating the HPA axis to induce anovulation and amenorrhoea (Lateef and Akintubosun, 2020). These effects are also seen in monkey models where a change of social environment or disruptive social interactions with group members can generate amenorrhoea (Adams et al., 1985; Bethea et al., 2005, 2008; Michopoulos et al., 2009; Wagenmaker et al., 2009).

Traumatic events such as sexual assault, incarceration or natural disasters may induce PTSD (Leeners et al., 2007; Beaglehole et al., 2018), which is another risk factor for FHA (Berga and Girton, 1989). For example, FHA has been reported after floods (Neuberg et al., 1999) and in women interned in a concentration camp during the Second World War before any malnutrition became evident (Sydenham, 1946). Incarcerated women show a higher prevalence of amenorrhoea, and women reporting additional stress factors such as childhood physical or sexual abuse, economic deprivation or coming from a racial and ethnic minority are also at increased risk (Allsworth et al., 2007).

It is also evident that behaviours such as over-exercise or restricting eating may reflect an underlying mental or psychiatric disease (Giles and Berga, 1993; Marcus et al., 2001; Berga, 2008). For instance, amenorrhoeic runners were found to suffer significantly more often from major affective or eating disorders than menstruating runners (Gadpaille et al., 1987). Also, women with anxiety disorders or depression have been found to be at increased risk for FHA (Fava et al., 1984; Joffe et al., 2006; Lawson et al., 2009).

Women with FHA often display combinations of psychological factors that can activate stress responses, induce metabolic disturbances and/or result in excessive exercising (Giles and Berga, 1993; Marcus et al., 2001; Faust et al., 2013). In rhesus monkeys, the combination of low-level psychosocial stress and moderate energy imbalance resulted in a higher proportion of abnormally long or anovulatory cycles than either stressor alone (Williams et al., 2001, 2007). Thus, psychological, psychogenic and metabolic stressors act synergistically in compromising reproduction (Giles and Berga, 1993; Berga et al., 1997; Warren et al., 1999; Marcus et al., 2001; Williams et al., 2007; Berga, 2008). In line with these findings, severe mental or infectious diseases such as HIV or Ebola infection have been found to add to the risk for FHA (Fava et al., 1984; Cejtin et al., 2018; Godwin et al., 2019). On this background, the current SARS-CoV-2 pandemic will probably also influence the prevalence of FHA.

Substantial investigation has focused on examining the mechanisms through which immune, nutritional and psychological stressors impact upon the reproductive axis (Li and O'Byrne, 2015). Stress activates multiple neural axes and results in elevated corticosteroid and prolactin concentrations. All of these factors, in turn, can impact upon the functioning of the GnRH neuronal network (Berga and Girton, 1989; Dobson et al., 2003; Mastorakos et al., 2006; Williams et al., 2007; Meczekalski et al., 2008; Li and O'Byrne, 2015). In many cases, it is difficult to tease apart the impacts of nutritional and psychological stressors on fertility with a synergism between the two in operation (Shanan et al., 1965; Fioroni et al., 1994; Berga, 1997; Chand and Lovejoy, 2011; Mendelson, 2013).

Cortisol. High cortisol levels are well-known to be associated with amenorrhoea (Suh et al., 1988; Berga and Girton, 1989; Berga et al., 1997; Brundu et al., 2006): Athletes with FHA as well as those with eating disorders and amenorrhoea have higher serum cortisol levels than women with a menstrual cycle (Villanueva et al., 1986; Laughlin et al., 1998; Ackerman et al., 2012). Administration of hydrocortisone reduces LH pulse frequency during the follicular phase of otherwise eumenorrhoeic women (Saketos et al., 1993). Elevated circulating corticosteroids reduce LH pulse amplitude (Olster and Ferin, 1987; Petraglia e al., 1987; Saketos et al., 1993; Dudas and Merchenthaler, 2002; Breen et al., 2008) and suppress LH pulse frequency through an unknown central mechanism (Ralph et al., 2016). Traumatic family events (sexual abuse, parental conflict, separation or death) have been associated with elevated cortisol levels that can persist beyond the traumatic period (Flinn et al., 2011; Jacobs et al., 2015) and consequently present a long-term risk factor for FHA.

Cortisol seems to be one of the most important factors in return of menses. In women with a history of eating disorders, and after weight gain, serum levels of fasting cortisol tend to be lower in women with return of menses compared to those without (Tinahones et al., 2005; Misra et al., 2006; Dei et al., 2008; Arimura et al., 2010; Pitts et al., 2014; Karountzos et al., 2017). Other studies confirmed significant differences in cortisol levels between women with and without return of menses (Miller et al., 1998; Falsetti et al., 2002; Jacoangeli et al., 2006) and two studies identified low serum cortisol levels as a predictor for the return of menses (Falsetti et al., 2002; Arimura et al., 2010). Furthermore, women with a return of menses after cognitive behavioural therapy had cortisol levels comparable to those of eumenorrhoeic women (Berga et al., 1997).

Corticotrophin releasing hormone. Corticotrophin-releasing hormone (CRH) suppresses pulsatile LH secretion but the pathway through which activated CRH neurons inhibit GnRH secretion remains unclear (Li and O'Byrne, 2015; McCosh et al., 2019). There is little evidence for a direct modulation of GnRH neurons by CRH although effects of CRH on kisspeptin neurons may exist (Raftogianni et al., 2018; McCosh et al., 2019).

Prolactin. Prolactin is released in response to stress in humans and animals (Schedlowski et al., 1992; Theorell, 1992; Armario et al., 1996; Sobrinho, 2003; Sonino et al., 2004; Levin et al., 2018). Animal studies show that prolactin acts through kisspeptin neurons to suppress pulsatile GnRH secretion, and infertility due to hyperprolactinaemia can be reversed to some extent by treatment with kisspeptin (Sonigo et al., 2012). Prolactin also raises secretion of adrenocorticotrophic hormone (ACTH) and augments the sensitivity of the adrenal cortex to ACTH, thus resulting in high corticosterone release even with low levels of ACTH (Weber and Calogero, 1991).

Adrenergic pathways. Acute stress results in the elevated secretion of catecholamines from the adrenal medulla and the activation of brainstem adrenergic neurons (Tank and Lee Wong, 2014). The effects of adrenaline and noradrenaline on LH secretion have been known for many decades (Sawyer, 1975) but the precise mechanisms through which they modulate GnRH secretion remain unclear (Herbison, 2015). In general, adrenergic inputs to the GnRH neuronal network are thought to exert a permissive role in enabling the network to function optimally for both pulsatile and surge secretion (Gallo et al., 1989; Scott and Clarke, 1993; Goodman et al., 1995; Anselmo-Franci et al., 1997; Herbison, 1997). Nevertheless, in gonadectomised animals, noradrenaline consistently suppresses pulsatile LH secretion (Herbison, 1997). Electrophysiological studies in mice have revealed that the activation of adrenergic receptors on GnRH neurons was exclusively inhibitory (Han and Herbison, 2008). Thus, it seems likely that the brainstem adrenergic neurons innervate multiple components of the GnRH neuronal network where they may exert different effects depending upon their level of activation (Herbison, 1997). For example, under states of heightened activation, they may provide a predominant direct inhibition of GnRH neurons to aid in the suppression of pulsatile LH secretion.

Serotonin. The likelihood of occurrence of FHA and return of menses may also be influenced by an individual’s sensitivity to stress. Evidence in humans and monkeys indicates that alterations in brain serotonin transmission play a role in determining an individual’s sensitivity to stress (Tancer et al., 1994; Ressler and Nemeroff, 2000; Bhagwagar et al., 2002; Bethea et al., 2008). Stress-sensitive monkeys were found to have diminished serotonergic activity and administration of a serotonin re-uptake inhibitor (citalopram) improves stress resilience (Bethea et al., 2005; Lima et al., 2009). Human and animal studies have also indicated that a serotonin transporter gene variant is involved in individual stress susceptibility and related suppression of LH secretion (Caspi et al., 2003; Grabe et al., 2005; Michopoulos et al., 2009; Caspi et al., 2010). Notably, a decrease in ovarian steroid hormone concentrations was found to suppress serotonin neural function in monkeys (Bethea et al., 2002; 2005) but it remains unclear to what extent circulating steroid levels are involved in maintaining FHA in women. Women with greater stress resilience show a reduced risk for irregular menstrual cycles when experiencing low to moderate chronic stress (Palm-Fischbacher and Ehlert, 2014). However, no studies evaluating stress-resilience in the context of resumption of menses have been conducted so far.

Lifestyle factors

Cigarette smoking is associated with higher rates of menstrual disorders (Howe et al., 1985; Buck et al., 1997). Several biologic mechanisms have been proposed to underlie this, such as actions on the hypothalamic-pituitary-ovarian axis, a direct toxic effect on the ovary and alterations in peripheral oestrogen production (Weisberg, 1985). In a study of 2544 college students, cigarette smoking showed a dose-dependent relationship with the risk for FHA (Johnson and Whitaker, 1992). Smoking more than one packet of cigarettes per day was associated with a 1.96 increased relative risk for persisting amenorrhoea. Unfortunately, no information was available on the lifetime duration of smoking. The well-established association between smoking and lower BMI (Rasky et al., 1996) is a potential confounder but student smokers in that study had the same relative weight as non-smokers.

The effect of alcohol intake on the menstrual cycle has not been clearly established (Mello, 1988; Grodstein et al., 1994). Recent moderate alcohol intake does not appear to have adverse short-term effects on menstrual cycle function (Schliep et al., 2015; Shilaih et al., 2017). The only study investigating alcohol intake in the context of prolonged amenorrhoea showed no association between alcohol consumption and persistence of FHA (Johnson and Whitaker, 1992). These findings have to be interpreted with caution as alcohol intake was self-reported.

Genetic factors

Genetic factors influence age at menarche (Dvornyk and Waqar ul, 2012) and at menopause (Voorhuis et al., 2010) and will very likely also be involved in the regulation of the menstrual cycle. Rare gene variants associated with idiopathic hypogonadotropic hypogonadism in women with hypothalamic amenorrhoea may influence susceptibility to functional changes in GnRH secretion (Caronia et al., 2011). Also, in comparison to regularly menstruating runners, runners with FHA reported more eating disorders or major affective disorders in close kinship (Gadpaille et al., 1987).

Treatment options trialled for re-onset and maintenance of menstruation

Medical treatment options for facilitating the recovery of menses include gonadal hormonal therapy (Falsetti et al., 2002; Genazzani et al., 2012; Shen et al., 2013), recombinant human leptin (metreleptin) (Welt et al., 2004; Chou et al., 2011) or opioid receptor blockers (naltrexone) (Genazzani et al., 1995). Altogether, six prospective studies are available (three of low quality, one of middle quality and two of high quality).

Medication

Oestrogen. The association between oestrogens and female weight has been previously reported (Leeners et al., 2017). An increase in oestradiol levels resulting from follicular maturation and conversion of androgens to oestrogens in fat tissue (associated with weight gain) is confirmed in all of the studies investigating hormonal levels in resumption of menses (Audi et al., 1998; Holtkamp et al., 2003; Jacoangeli et al., 2006; Arimura et al., 2010; Cominato et al., 2014; Pitts et al., 2014; Barakat et al., 2016; Karountzos et al., 2017; Tokatly Latzer et al., 2019). Oestrogen increases GnRH receptor gene expression and enhances the ability of FSH to induce expression of LH receptors and promote follicular growth (Richards et al., 1976; Turzillo et al., 1998). This has generated the hypothesis that the pituitary response to GnRH and the ovarian response to gonadotropin could be reinforced with oestrogen. However, studies evaluating this hypothesis failed to show any effect on the likelihood of recovery of menses (Falsetti et al., 2002; Genazzani et al., 2012; Shen et al., 2013).

Metreleptin. Recombinant human leptin (metreleptin) facilitated the recovery of menses in women with FHA due to low body weight, excessive exercise or unspecified reasons (Welt et al., 2004; Chou et al., 2011). With two daily subcutaneous doses (0.08–0.12 mg/kg body weight) to mimic normal diurnal patterns, recovery of menses occurred after 28 days up to 32 weeks (Welt et al., 2004; Wong et al., 2004; Chou et al., 2011). In women, without return of menses during the study period, continuous improvement of follicular maturation was confirmed by ultrasound and laboratory parameters (Welt et al., 2004). Important adverse effects of metreleptin application include a decrease in appetite that is counter-productive to energy status (Welt et al., 2004; Chou et al., 2011). Although results are promising, further investigations are necessary to determine the efficacy and safety of metreleptin treatment including dose-finding and treatment duration studies with different background characteristics.

Naltrexone. The long-acting opioid receptor blocker naloxone increases LH pulse frequency and amplitude in amenorrhoeic women (Quigley et al., 1980; Khoury et al., 1987) and, as such, naltrexone has been tested to induce the recovery of menses. Administration of 50–150 mg/day was found to cause no effect (Remorgida et al., 1990), a moderate effect (Armeanu et al., 1992) or full return of menses (Wildt and Leyendecker, 1987; Wildt et al., 1993). In the most recent placebo-controlled study, 24 (80%) women with FHA due to weight loss reported menstrual bleeding within 90 days after initiation of naltrexone therapy (Genazzani et al., 1995). After 3 months of treatment, LH plasma levels and pulse amplitude increased whereas FSH plasma levels did not show any change. The recovery of menstrual cycles occurred prior to weight gain, suggesting that naltrexone had a central effect independent from body weight gain. Six months after naltrexone discontinuation, 75% of the women were still eumenorrhoeic. Although only minor side effects such as nausea at the beginning of treatment have been reported, further studies are necessary to determine efficacy and safety in a larger sample of women with FHA distinguished by its trigger.

Psychotherapeutic interventions

Women with FHA, especially those with eating disorders, often experience fear of weight gain, concerns about dieting, weight judgements of others, perfectionistic performance standards, tendencies to engage in binge eating, excessive exercise or depressive symptoms (Fries, et al., 1974; Giles and Berga, 1993; Marcus et al., 2001; Brambilla et al., 2003; Favaro and Santonastaso, 2009; Faust et al., 2013). Many of these characteristics can effectively be treated by CBT (Berga et al., 2003; Berga and Loucks, 2006; Michopoulos et al., 2013). CBT may successfully modify eating and exercise patterns as well as maladaptive attitudes concerning body image, weight regulation and problem-solving techniques (Giles and Berga, 1993; Marcus et al., 2001; Pauli and Berga, 2010; Mountjoy et al., 2018). CBT has been shown to promote the return of menses alongside the recovery of cortisol or leptin levels (Berga et al., 1997, 2003; Berga and Loucks, 2006; Michopoulos et al., 2013).

Besides CBT, hypnotherapy may be helpful in women with FHA (Tschugguel and Berga, 2003). Coping skills such as relaxation, distress tolerance, meditation, mindfulness and yoga have provided further positive effects (Berkman et al., 2006; Rani et al., 2011; Goyal et al., 2014; Katterman et al., 2014; Hall et al., 2016), but have not directly been investigated in their effect on recovery of menses.

Discussion

While disturbances of the menstrual cycle in association with weight loss and reduced energy balance are relatively well understood, knowledge on which conditions and whether menses recovers after FHA is rather limited. Recently published guidelines of the Endocrine Society provide well-founded recommendations for the diagnosis of FHA as well as the treatment of its consequences such as bone loss, infertility or eventual cardiovascular impairment (Gordon et al., 2017). The present review focusses on the return of menses, i.e. it expands these guidelines into the physiology of recurrence of the menstrual cycle and options to end FHA.

Providing a comprehensive account of the return of menses after FHA is challenged by the many methodological differences used in study designs examining this issue and the difficulties in teasing out the impact of the influencing factors. Research groups use various definitions of FHA and also of the return of menses; for example, with regard to the total number of cycles requested, the regularity of cycles or the presence of ovulation. Only 11 of the 31 studies exploring recovery of menses after FHA presented a definition of the return of menses and often the duration of initial amenorrhoea was not reported. Most studies do not differentiate between full ovulatory cycles and partial recovery.

In addition, the conditions to be met to initiate the observation phase (i.e. body weight/BMI, energy intake/expenditure, etc.) when return of menses can justifiably be expected, are often either not defined, not confirmed or not controlled for stability during the observation phase. A further important point is the marked variation in the total length of the follow-up period.

We also note that many different approaches have been used to define and evaluate influencing factors such as measures of weight/body fat, energy balance, nutrition, psychological and lifestyle factors. Often, studies have not controlled for well-known hormonal causes of cycle disturbances and endocrine diseases. This is particularly important because, at present, we cannot distinguish between causes and consequences for most gut hormones or other laboratory parameters. Differences in study groups related to causes of FHA, age, ethnicity and psychological background as well as small sample sizes further challenge the generalizability of findings.

Only 12 studies had a prospective design and currently, there are no prospective longitudinal studies with adequately powered study groups that control for relevant confounders to allow a reliable estimation of the frequency and timing of recovery of menses. Consequently, our knowledge of the prevalence of recovery of menses comes from studies that are of a very limited quality. Today, we cannot identify women with an increased risk for long-term FHA even after weight/BMI and energy availability have been normalised. It is very likely that psychological factors and stress are involved in those women returning to normal weight without a return of menses (Gordon et al., 2017), but presently no studies on success rates of treatment concepts including both organic and psychological factors are available. Therefore, we cannot as yet answer the question of whether the return of menses can be reliably achieved with adequate treatment.

While there is abundant research on the onset of the cycle during puberty and the initiation of amenorrhoea in the case of weight loss or unfavourable energy balance, literature on the physiology of recovery of menses is virtually non-existent. Animal studies have allowed an understanding of the regulation of the pulse generator and how different gut hormones interact with the GnRH neuronal network, but more research is needed to better understand the regulatory mechanisms involved in the recovery of menses in women.

Although study designs vary considerably with regard to parameters and techniques chosen, research findings consistently support the importance of maintaining of a stable normal weight, a minimum amount of body fat and adequate energy availability for recovery of menses. However, we lack a reliable basis to suggest target weights or energy balance that would allow the recovery of menses in individual women. Also, we cannot predict whether the menstrual cycle recovers after the normalisation of weight and energy balance in individual women and how lifestyle or psychological factors are involved. Studies examining the association between sport and return of menses show higher chances of recovery in normal weight women when energy availability increases in most but not in all studies. The small number of study participants, self-reported food intake (known to be biased by underreporting) and limited precision of the evaluation of energy balance might explain differences in findings.

The calculation of nutritional needs based on total energy expenditure and sufficient carbohydrate and protein requirements derived from individual goals and sport-specific regimens is the major factor promoting the return of menses in athletes (Kopp-Woodroffe et al., 1999; Arends et al., 2012; Mallinson et al., 2013; Cialdella-Kam et al., 2014; Lagowska et al., 2014; Reed et al., 2015). In contrast, weight recovery is the most important factor in women with FHA due to eating disorders (Golden et al., 1997, 2008; Dempfle et al., 2013; Faust et al., 2013). While an indirect effect of psychological factors especially in the context of eating disorders is well-established, the direct effect of previous and current psychological factors warrants further exploration. Very likely, subtle persisting characteristics of eating behaviours, exercise and energy balance are involved in augmenting the risk for prolonged amenorrhoea but these remain to be elucidated.

Overall, the time to return of menses seems to depend on the initial cause of FHA with shorter duration till recovery in athletes (Cialdella-Kam et al., 2014) compared to women with eating disorders (Karountzos et al., 2017). The inhomogeneity of studies and the lack of systemic consideration of influencing factors hamper development of good and robust prediction models.

Comparative studies or reviews on the effect of different treatment strategies to improve chances for return of menses do not exist. Interventions such as increase in weight/energy availability or adaptation of nutrition have been confirmed to facilitate the return of menstruation; however, specific nutrition concepts have only been investigated in athletes. Stress management is part of the therapeutic approach with eating disorders, but its role in improving the chances of return of menses has not yet been fully explored. In addition to the well-known strategies of adequate weight/BMI/percentage body fat, energy balance and psychological well-being, drugs such as metreleptin or naltrexone have showed favourable results in some studies but need further investigation.

Open research questions

More methodologically well-designed studies are needed to close the present gap in our understanding on how the return of menses in FHA can be facilitated. This includes clear definitions and diagnostic criteria of FHA, return of menses, conditions at the beginning of an observation period, interventions and influencing factors/potential confounders. Further research should address clinically relevant therapeutic target values for weight, BMI and energy levels as well as overall chances of return of menses. Studies should be prospective and longitudinal and include control groups with medications and other therapeutic interventions tested in a placebo-controlled double-blind design, wherever possible. The pathophysiology of return of menses should also be evaluated in human studies, and should cover central and peripheral regulatory mechanisms, actual psychological stressors and early stress experiences, as well as genetic and epigenetic factors involved in biological and psychological influences. The latter could serve to identify women at risk of FHA and enable preventive measures.

Practical advice

Clinicians should inform women with FHA on the potential long-term consequences of amenorrhoea and the likelihood of return of menses under adequate as well as inadequate conditions. Extensive counselling may be needed in the case of psychopathology or in high-performance athletes. As FHA is often the combined effect of low weight, excessive exercise, poor nutrition and psychological factors, evaluation of the cause of amenorrhoea should always include the exploration of all relevant factors and be carefully differentiated from organic causes of amenorrhoea. FHA can persist for several years even after achieving a healthy weight/adequate energy balance and has severe long-term effects if not treated. As such, it is important to aim as early as possible for normal cycle function. Therapeutic support should include adequate nutrition, energy balance, exercise level and mental health. Ideally, support should be provided by an interdisciplinary team of a gynaecologist–obstetrician, a dietitian and a mental health professional. In particular, behavioural changes needed for the improvement of energy balance may necessitate psychotherapeutic support. Non-drug treatment options that eliminate the causes of FHA remain the therapy of choice to regain menstruation. Predicting if and when menses will return is difficult. As the chances of pregnancy strongly rely on women’s age (Leeners et al., 2013; Somigliana et al., 2016), medically assisted reproduction support should be discussed when menses do not reoccur despite adequate nutrition, weight, energy balance and stress management.

Conclusion

Based on the limited evidence available, the restoration of adequate weight/BMI and energy balance have a clear beneficial effect, but normalisation of these factors alone does not reliably result in return of menses. As the physiology of the return of menses and the underlying factors are only poorly understood, it seems unlikely that a precise prediction model for when the menstrual cycle will recover in individual women will be achieved in the near future. Given the severe impact of FHA on women’s overall and reproductive health, there is a clear and urgent need for further research investigating the factors allowing return of menses. Currently, a combination of adequate nutrition, weight, energy balance and stress management still seems to be the most adequate approach to increase the chances of the recovery of menses after FHA.

Authors’ roles

J.P and B.L. contributed to the identification and critical evaluation of the relevant literature, the analysis of study results and the drafting the article including the critical discussion of findings. J.P. participated in preparation of the proposal, completed the initial literature research, drafted a first version of the article and participated in finalisation of the article. A.H. provided his expertise on central regulatory mechanisms in return of menses, drafted and finalised related passages of the manuscript and critically revised the final version of the manuscript. B.L. finalised the concept for the manuscript and supervised the literature research, data extraction and presentation of relevant data. B.L. drafted and finalised different versions of the manuscript and critically revised the final version of the manuscript. All authors approved the final version of this manuscript.

Funding

This study did not receive any specific funding.

Conflicts of interest

None of the authors has any conflict of interest.