Long-term outcome of prophylactic thyroidectomy in children carrying RET germline mutations

Abstract

Background

A comprehensive assessment has not been undertaken of long-term outcomes in children carrying germline RET mutations and undergoing prophylactic thyroidectomy with the aim of preventing medullary thyroid cancer (MTC).

Methods

A retrospective outcome study (1994–2017) of prophylactic thyroidectomy in children, with and without central node dissection, was performed at a tertiary surgical centre.

Results

Some 167 children underwent prophylactic thyroidectomy, 109 without and 58 with concomitant central node dissection. In the highest-risk mutational category, MTC was found in five of six children (83 per cent) aged 3 years or less. In the high-risk category, MTC was present in six of 20 children (30 per cent) aged 3 years or less, 16 of 36 (44 per cent) aged 4–6 years, and 11 of 16 (69 per cent) aged 7–12 years (P = 0·081). In the moderate-risk category, MTC was seen in one of nine children (11 per cent) aged 3 years or less, one of 26 (4 per cent) aged 4–6 years, three of 26 (12 per cent) aged 7–12 years, and seven of 16 (44 per cent) aged 13–18 years (P = 0·006). Postoperative hypoparathyroidism was more frequent in older children (32 per cent in the oldest age group versus 3 per cent in the youngest; P = 0·002), whether or not central node dissection was carried out. Three children developed recurrent laryngeal nerve palsy; all had undergone central node dissection (P = 0·040). All complications resolved within 6 months. Postoperative normalization of calcitonin serum levels was achieved in 114 (99·1 per cent) of 115 children with raised preoperative values. No residual structural disease or recurrence was observed.

Conclusion

Early prophylactic thyroidectomy is a viable surgical concept in experienced hands, sparing older children the postoperative morbidity associated with delayed neck surgery.

Introduction

The widespread availability of more sensitive calcitonin assays has changed the scene for hereditary medullary thyroid cancer (MTC), a neuroendocrine malignancy dubbed C-cell cancer in reference to its ability to synthesize and secrete calcitonin. In the pregenomic era, biochemical screening programmes led to earlier selection of children for prophylactic thyroidectomy and to a decrease in primary tumour size from 0·8 to 0·2 cm. The proportion of bilateral neoplasms declined from 100 to 13 per cent, and the rate of lymph node metastasis reduced from 58 to 0 per cent¹. Because affected children remained free from MTC for a mean of 11 years after thyroidectomy, regular biochemical screening and early thyroidectomy were thought to prevent metastasis and death from hereditary MTC². Biochemical (calcitonin) screening was also shown to be more sensitive than anatomical screening such as neck ultrasonography³.

The advent of the genomic era 25 years ago further enhanced this shift towards smaller tumours, augmenting tumour lead time. Lead time is defined as the length of time between tumour detection through screening and the usual clinical manifestation of the disease. With the ground-breaking discovery of RET (REarranged during Transfection) as the susceptibility gene on chromosome 10q11.2 in 1993^4,5, it became immediately apparent that DNA-based screening, disclosing an individual's genetic predisposition to MTC, was superior to biochemical screening for subclinical disease^6–9. Genomic confirmation of this hereditable predisposition enabled truly prophylactic thyroidectomies in infants and young children who still had normal thyroid glands or no more than neoplastic C-cell hyperplasia^10,11.

Earlier identification of asymptomatic infants and young children as gene carriers opened a window of opportunity, within which total thyroidectomy without lymph node dissection represents adequate therapy^12,13. Complete removal of this expendable target organ eliminates once and for all the potential for malignant transformation inherited by every thyroid C-cell, while sparing the child the incremental morbidity attendant to central lymph node dissection, specifically postoperative hypoparathyroidism¹⁴. The goal is not so much to prevent malignancy from occurring in the first place, but to remove the thyroid before metastasis develops¹⁵. In children with raised preoperative calcitonin serum levels, the result of the prophylactic thyroidectomy is evaluated from the normalization of postoperative basal calcitonin serum levels.

Although already 25 years into the genomic era, a comprehensive appraisal of the long-term outcomes following prophylactic thyroidectomy, with or without central node dissection, has not been performed in children carrying RET germline mutations, weighing mutation-specific, age-dependent tumour progression and clinical outcome against age-related operative morbidity. The aim of this retrospective cohort study, drawing on the largest series of paediatric RET gene carriers reported to date, was undertaken to fill this void.

Methods

Study cohort

All children (aged 18 years or less) carrying a RET germline mutation in exons 10, 11, 13, 14, 15 or 16 of the RET proto-oncogene and referred for prophylactic neck surgery between November 1994 and June 2017 were included. These children had serum calcium and calcitonin levels determined immediately before thyroidectomy, upon discharge, and subsequently at the discretion of the attending physicians elsewhere. Informed consent was obtained from all patients and/or their legal guardians before the operation, in line with standard practice of care according to the practice guidelines of the German Society of Surgery¹⁶. Node dissection was performed more often in the beginning because initial genetic association studies seemingly implied more rapid progression of MTC¹⁷. In line with the 2015 revised American Thyroid Association (ATA) practice guidelines for the management of MTC¹⁸, the term prophylactic thyroidectomy is used here to define removal of the thyroid before MTC develops or while it is clinically inapparent and confined to the gland. Retrospective analysis of existing data sets from routine patient care does not require institutional review board approval under national law and applicable institutional regulations.

Protection of recurrent laryngeal nerves, parathyroid glands and thymus

All surgical procedures were performed using optical magnification and bipolar coagulation, as detailed elsewhere¹⁹. Great care was taken to identify and preserve the recurrent laryngeal nerves and parathyroid glands¹⁴. After December 1997, all children and adults underwent thyroidectomy with the use of intraoperative nerve monitoring as an institutional standard of care²⁰. Every reasonable effort was made in infants and young children to leave the thymus and normal parathyroid glands in situ with good blood supply and, where not feasible, to autotransplant parathyroid slivers.

Histopathological diagnosis of medullary thyroid cancer

All surgical specimens were subjected to histopathological examination including immunostaining for calcitonin. MTC was diagnosed: in the presence of tumour extension beyond the basement membrane; on demonstration of lymphatic or vascular invasion on histopathology; or any combination of the above. Primary tumour diameter was measured directly on the surgical thyroid specimens. A diagnosis of nodal metastasis required histopathological confirmation.

Outcome measurements

Each child's vocal fold function was determined by laryngoscopy before and after the operation. If abnormal, laryngoscopic examinations were repeated to confirm recovery of the recurrent laryngeal nerve.

When signs and symptoms of early postoperative hypoparathyroidism were present, patients were started on calcium and vitamin D replacement therapy for 2 weeks and then instructed to taper off as permitted by clinical symptoms and serum levels of calcium and intact parathyroid hormone (PTH).

Outcome information from other institutions, general practitioners, and/or the patients or their parents was collected as necessary. Special emphasis was placed on postoperative vocal fold status based on laryngoscopic examination, current need for calcium and vitamin D replacements, and the most recent serum levels of intact PTH.

Any vocal fold palsies or calcium and vitamin D replacements persisting for more than 6 months were considered as permanent vocal fold palsy and postoperative hypoparathyroidism respectively.

Basal and stimulated serum calcitonin levels were regarded as normal, both before and after the operation, when the upper normal limit of the calcitonin assay used was not exceeded.

Statistical analysis

Categorical and continuous data were tested on univariable analysis with the two-tailed Fisher's exact test and the two-tailed Mann–Whitney–Wilcoxon rank sum test respectively. Children were stratified by age using bands of 3 years or less, 4–6, 7–12 and 13–18 years, ensuring sufficient numbers of children and events in each group. To assess trends in categorical variables by age group, the linear-by-linear association χ² statistic was calculated. Receiver operating characteristic (ROC) curve analysis was used to identify age thresholds of increased operative morbidity. Multiple testing was adjusted for with the Bonferroni correction²¹. The software package SPSS® version 24 (IBM, Armonk, New York, USA) was used for statistical analysis. The level of significance was set at P < 0·050.

Results

Study cohort

In total, 183 children were identified who underwent total thyroidectomy without (109 children) or with (74 children, of whom 16 also had lateral node dissection) central node dissection. Because they presented with suspicious nodes, the 16 children with lateral in addition to central node dissection (12 node-positive and 4 node-negative children) were not included in the analysis of prophylactic thyroidectomy. These 16 children carried mutations in codon 918 (10 node-positive children aged 6–17 years, with a median of 35·5 (range 1–68) neck metastases, and 3 node-negative children aged 1·5, 3 and 14 years), codon 634 (a 17-year-old girl with 3 node metastases, and 3 node-negative children aged 6, 14 and 16 years) and codon 790 (a 10-year-old boy with 20 node metastases).

The remaining 167 children (109 without and 58 with central but not lateral node dissection) formed the study cohort.

Genetics of the study cohort

The 167 children in the study cohort had RET mutations in codon 634 (73 children; 43·7 per cent), codon 618 (17 children; 10·2 per cent), codon 620 (16 children; 9·6 per cent), codon 790 (12 children; 7·2 per cent), codons 611, 791 and 891 (9 children each; 5·4 per cent), codon 804 (8 children; 4·8 per cent), codon 918 (6 children; 3·6 per cent), codon 768 (3 children; 1·8 per cent), codons 630 and 852 (2 children each; 1·2 per cent) and codon 609 (1 child; 0·6 per cent).

Based on their malignant transformation properties, these 13 RET mutation types can be grouped, reflecting the 2015 revised ATA practice guidelines¹⁸ and more recent evidence^22,23, as follows (Fig. 1): six children (3·6 per cent) into the highest (HST) risk ATA category (M918T); 73 (43·7 per cent) into the high (H) risk ATA category (C634x); 77 (46·1 per cent) into the moderate (MOD) risk ATA category (C609x, C611x, C618x, C620x, C630x; E768D, L790F, V804L/M and S891A); and 11 children (6·6 per cent) into the variant of unknown significance (VUS)/polymorphism (PM) category (Y791F and I852M).

Fig. 1

RET (REarranged during Transfection) tyrosine kinase receptor genotype and clinical implications. *Based on recent evidence^22,23, Y791F and I852M RET variants no longer qualify as mutations. ATA, American Thyroid Association; CLD, cadherin-like domain; b+stCT, preoperative basal and stimulated serum calcitonin levels; WNL, within normal limits; MOD, moderate; CRD, cysteine-rich domain; H, high; TM, transmembrane domain; JM, juxtamembrane domain; TK, tyrosine kinase domain; HST, highest; ASAP, as soon as possible

Demographic, genetic and histopathological composition of the study cohort

Table 1 depicts the demographic genetic, and histopathological characteristics of all 167 children who underwent prophylactic thyroidectomy, stratified by the presence or absence of central node dissection. Children had concomitant central node dissection more often in the earlier years of the study (median operative year 2003 versus 2007; P < 0·001), reflecting the state of knowledge at the time¹⁷. Likewise, these children tended to be slightly older at prophylactic thyroidectomy (median age 7 versus 5 years; P = 0·015) and more often to have occult MTC (41 versus 23·9 per cent; P = 0·022), even though statistical significance was lost for these associations after correction for multiple testing (Table 1).

Table 1

Demographic, genetic and histopathological characteristics of the cohort

		Prophylactic thyroidectomy
	Total (n = 167)	Without central neck dissection (n = 109)	With central neck dissection (n = 58)	P†
Age at prophylactic thyroidectomy (years)				0·015‡
Mean (95 per cent c.i.)	6·8 (6·1, 7·4)	6·2 (5·4, 6·9)	7·9 (6·6, 9·1)
Median (range)	6 (0·5–18)	5 (0·5–18)	7 (0·5–18)
Sex ratio (M : F)	78 : 89	49 : 60	29 : 29	0·630
Updated 2015 ATA risk category18				0·076
Highest (HST)	6 (3·6)	2 (1·8)	4 (7)
High (H)	73 (43·7)	43 (39·4)	30 (52)
Moderate (MOD)	77 (46·1)	57 (52·3)	20 (34)
Polymorphism*	11 (6·6)	7 (6·4)	4 (7)
Histology				0·022
Normal/CCH	117 (70·1)	83 (76·1)	34 (59)
MTC N0	50 (29·9)	26 (23·9)	24 (41)
MTC N1	0 (0)	0 (0)	0 (0)
Year of prophylactic thyroidectomy				< 0·001‡§
Mean (95 per cent c.i.)	2005 (2004, 2006)	2006 (2005, 2008)	2003 (2001, 2004)
Median (range)	2005 (1994–2017)	2007 (1995–2017)	2003 (1994–2017)

Values in parentheses are percentages unless indicated otherwise.

*

Includes only Y791F and I852M variants, which, based on recent evidence^22,23, no longer qualify as MOD mutations as listed in the 2015 American Thyroid Association (ATA) guideline. CCH, C-cell hyperplasia; MTC, medullary thyroid cancer.

†

Fisher's exact test, except ‡Mann–Whitney–Wilcoxon rank sum test;

§

statistically significant after Bonferroni correction for multiple testing.

Prevalence of medullary thyroid cancer by age and mutational risk profile

Table 2 specifies the age-dependent malignant transformation to early node-negative MTC for mutational risk categories. In the highest-risk mutational group (ATA category HST), MTC was present in five (83 per cent) of six children aged 3 years or less. All children older than 3 years in this group needed lateral node dissection and hence were excluded from the study. In the high-risk mutational group (ATA category H), MTC was encountered in six (30 per cent) of 20 children aged 3 years or less, 16 (44 per cent) of 36 children aged 4–6 years, and 11 (69 per cent) of 16 children aged 7–12 years (P for trend = 0·081). Children older than 12 years almost always required additional lateral node dissection, and thus were not included in the study. In the moderate-risk mutational group (ATA category MOD), MTC was found in one (11 per cent) of nine children aged 3 years or less, one (4 per cent) of 26 children aged 4–6 years, three (12 per cent) of 26 children aged 7–12 years, and seven (44 per cent) of 16 children aged 13–18 years (P for trend = 0·006). No MTC was noted in the 11 children carrying the Y791F or I852M sequence variants.

Table 2

Prevalence of medullary thyroid cancer by age and mutational risk

	Children with MTC
		Age (years)
Updated 2015 ATA risk category*	Total	≤ 3	4–6	7–12	13–18	P for trend‡
Highest (HST)	5 of 6 (83)	5 of 6 (83)	–	–	–	–
High (H)	33 of 73 (45)	6 of 20 (30)	16 of 36 (44)	11 of 16 (69)	[0 of 1]†	0·081
Moderate (MOD)*	12 of 77 (16)	1 of 9 (11)	1 of 26 (4)	3 of 26 (12)	7 of 16 (44)	0·006§
P for trend‡	< 0·001§	0·007§	< 0·001§	< 0·001§	–

Values in parentheses are percentages.

*

Based on recent evidence, Y791F and I852M RET variants no longer qualify as mutations (2015 American Thyroid Association (ATA) category MOD).

†

Censored. MTC, medullary thyroid cancer.

‡

Linear-by-linear association χ² statistic;

§

statistically significant after Bonferroni correction for multiple testing.

Within each age category, the pace of malignant progression to early node-negative MTC also differed according to genetic risk (Table 2). For children aged 3 years or less: from 83 per cent (5 of 6 children; ATA category HST) to 30 per cent (6 of 20 children; ATA category H) and 11 per cent (1 of 9 children; ATA category MOD) (P for trend = 0·007). For children aged 4–6 years: from 44 per cent (16 of 36 children; ATA category H) to 4 per cent (1 of 26 children; ATA category MOD) (P for trend < 0·001). For children aged 7–12 years: from 69 per cent (11 of 16 children; ATA category H) to 12 per cent (3 of 26 children; ATA category MOD) (P for trend < 0·001).

Overall, MTC was present in three (6 per cent) of 47 children with normal basal and stimulated calcitonin levels, in eight (24 per cent) of 33 children with normal basal and raised stimulated calcitonin levels, and in 39 (45 per cent) of 87 children with raised basal calcitonin levels (P < 0·001).

Surgical morbidity by age and extent of surgery

Transient postoperative hypoparathyroidism was increasingly more common in older children: 32 per cent in children aged 13–18 years versus 3 per cent in those aged 3 years or less (P for trend = 0·002) (Table 3). This trend was accentuated more in children who had central node dissection in addition to total thyroidectomy (33 versus 0 per cent respectively; P for trend = 0·031) compared with children undergoing total thyroidectomy alone (30 versus 4 per cent respectively; P for trend = 0·091). In ROC curve analysis, no clear-cut age thresholds could be discerned (data not shown). All 31 instances of transient postoperative hypoparathyroidism resolved within 6 months.

Table 3

Transient postoperative hypoparathyroidism by age and extent of surgery

	Children with transient postoperative hypoparathyroidism
		Age (years)
Extent of surgery	Total	≤ 3	4–6	7–12	13–18	P for trend*
Any surgery	31 of 167 (18·6)	1 of 35 (3)	11 of 64 (17)	13 of 49 (27)	6 of 19 (32)	0·002†
Prophylactic thyroidectomy alone	19 of 109 (17·4)	1 of 25 (4)	10 of 48 (21)	5 of 26 (19)	3 of 10 (30)	0·091
Prophylactic thyroidectomy with central node dissection	12 of 58 (21)	0 of 10 (0)	1 of 16 (6)	8 of 23 (35)	3 of 9 (33)	0·031
P for trend*	0·677	1·000	0·265	0·332	1·000

Values in parentheses are percentages.

*

Linear-by-linear association χ² statistic;

†

statistically significant after Bonferroni correction for multiple testing.

Transient recurrent laryngeal nerve palsy was noted in three children, all with central node dissection (P = 0·040) (Table 4). All three palsies resolved within 6 months.

Table 4

Transient recurrent laryngeal nerve palsy by age and extent of surgery

	Children with transient recurrent laryngeal nerve palsy
		Age (years)
Extent of surgery	Total	≤ 3	4–6	7–12	13–18	P for trend*
Any surgery	3 of 167 (1·8)	1 of 35 (3)	0 of 64 (0)	1 of 49 (2)	1 of 19 (5)	0·547
Prophylactic thyroidectomy alone	0 of 109 (0)	0 of 25 (0)	0 of 48 (0)	0 of 26 (0)	0 of 10 (0)	–
Prophylactic thyroidectomy with central node dissection	3 of 58 (5)	1 of 10 (10)	0 of 16 (0)	1 of 23 (4)	1 of 9 (11)	1·000
P for trend*	0·040	0·286	–	0·469	0·474

Values in parentheses are percentages.

*

Linear-by-linear association χ² statistic.

Wound infections were noted in two children who had prophylactic thyroidectomy without central node dissection: a 3-year-old girl who needed open wound revision on postoperative day 5, and a 7-year-old boy requiring intravenous antibiotics only.

Clinical outcome

Follow-up information after discharge from the hospital was available for 149 (89·2 per cent) of the 167 children: for 99 (90·8 per cent) of 109 children without and 50 (86 per cent) of 58 children with central node dissection (P = 0·016).

Postoperative normalization of calcitonin serum levels was achieved in 114 (99·1 per cent) of 115 children with raised preoperative calcitonin serum levels. The only child not cured biochemically was a 7-month-old boy with a basal calcitonin level of 105 pg/ml who carried the M918T mutation. Prophylactic thyroidectomy revealed a 2-mm large MTC confined to the right thyroid lobe. Basal calcitonin serum levels initially normalized only to rise again, reaching 68·6 pg/ml 10 months after thyroidectomy. No node metastases were identified despite meticulous neck dissection and histopathological examination, yielding 29 central and 72 lateral uninvolved nodes.

Within the median observational period of 72 (mean 84, range 1–255) months, none of the 149 children with follow-up information had evidence of residual structural disease or recurrence. Only one child died, from trauma unrelated to the RET mutation.

Discussion

This study describes the surgical outcomes of a large cohort of children carrying germline RET mutations who underwent prophylactic thyroidectomy to minimize the risk of later MTC. Younger children aged 3 years or less did not have greater surgical morbidity than older children. This makes a case for prophylactic thyroidectomy in affected infants and small children once calcitonin serum levels have increased.

Twenty-five years after the dawning of the genomic era, the age-related progression of MTC is well defined for hotspot mutations in codons 609, 611, 618, 620, 630, 634, 768, 790, 804, 891 and 918 at a population level. These close genotype–phenotype relationships form a robust genetic framework within which C-cell hyperplasia is transformed to MTC¹². At the individual level, genetic data alone, however, do not allow for accurate prediction of this transformation in time (age at onset). Biochemical (calcitonin) screening is needed and thus opens a window of opportunity for surgery, defined by the basal serum calcitonin level, which correlates with progression of the C-cell disease. As long as basal calcitonin serum levels remain within normal limits, MTC has not yet developed^12,13. When these levels cross the upper normal limit (usually less than 10 pg/ml for children aged 2 years or more²⁴), malignant transformation to node-negative MTC may have occurred. This time point may represent the last opportunity to perform prophylactic thyroidectomy without a need for central lymph node dissection.

Hereditary MTC is estimated to grow annually by 0·4–0·5 mm for node-negative tumours and 1·2–2·6 mm for node-positive tumours, involving 0·6–0·7 nodes per year²⁵. This slow tumour growth provides some latitude in affected children whose basal calcitonin serum levels remain within normal limits, so that prophylactic thyroidectomy can be postponed, but not avoided, for some time. As long as basal calcitonin serum levels remain at 30 pg/ml or below, node metastases are absent²⁶.

Biochemical cure, as indicated by postoperative normalization of the raised serum calcitonin levels, was accomplished here in 114 (99·1 per cent) of 115 children operated on at a mean age of 6 (median 5) years, compared with 44 (88 per cent) of 50 children who had surgery at a mean age of 10 years¹¹. These rates underline the importance of early prophylactic surgery in children carrying germline mutations of the RET gene. The result of the operation is, however, more difficult to determine in infants and young children who displayed normal rather than increased preoperative serum calcitonin levels.

Postoperative calcium and vitamin D replacement was administered significantly more often to older children (32 per cent of the oldest age group versus 3 per cent of the youngest; P = 0·002). This finding was remarkable because the parathyroid glands of infants and young children are typically smaller, more translucent and harder to differentiate from adjacent soft tissues, thymus and central neck nodes^27,28. On the other hand, older children have their thymus and lower parathyroid glands removed more often than younger children, possibly explaining the more frequent calcium and vitamin D replacements in older children. Because it can be challenging to replace calcium and vitamin D in infants and young children^15,28,29, the low rates of transient (1 of 35 children (3 per cent) aged 3 years or less) and permanent (0 per cent) hypoparathyroidism in the present cohort are noteworthy.

In patients undergoing prophylactic thyroidectomy without central node dissection, the transient and permanent hypoparathyroidism rates in the present study compare favourably against the sparse international literature on the subject: 4 per cent (1 of 25) and 0 per cent (0 of 25) for children aged 3 years or less operated on in 1994–2017, compared with 44 per cent (4 of 9) and 22 per cent (2 of 9) for children aged less than 3 years operated on in 1993–2014²⁹. The corresponding overall rates were 17·4 per cent (19 of 109) and 0 per cent (0 of 109) for children aged 18 years or less, compared with 27 per cent (12 of 44) and 20 per cent (9 of 44) in children less than 18 years old²⁹. In children who had prophylactic thyroidectomy with central node dissection, permanent postoperative hypoparathyroidism was seen in 6 per cent (3 of 50) in an earlier series¹¹, whereas it was zero (based on 149 children) in the present study.

This retrospective cohort study has many limitations that are common in studies of rare diseases. Findings from a high-volume tertiary referral centre dedicated to paediatric cancer surgery may not extend to other clinical settings. In addition, the 58 children with central lymph node dissection underwent prophylactic surgery slightly later (at a median age of 7 years versus 5 years for those without central node dissection; P = 0·015) and somewhat more frequently earlier in the study (median operative year 2003 versus 2007 respectively; P < 0·001). It also cannot be ruled out completely that older children could show and verbalize signs and symptoms of transient hypoparathyroidism more effectively than younger children, thus prompting more frequent calcium and vitamin D replacements in the immediate postoperative period. The overall effect of the age-specific differences in this study were, however, fairly subtle and did not translate into significantly greater operative morbidity. Because none of the 167 children operated on prophylactically had node metastases on histopathological examination, these findings also are limited to clinically node-negative children. Children with clinically apparent node metastases, typically in the context of multiple endocrine neoplasia type 2B or papillary thyroid cancer, require therapeutic dissection of enlarged nodes, which is associated with more frequent postoperative transient (34 per cent) and permanent (10 per cent) hypoparathyroidism¹⁴ than that following prophylactic node dissection.

Total thyroidectomy effectively eliminates the risk of persistent or recurrent C-cell disease in RET germline mutation gene carriers once and for all. As soon as calcitonin serum levels exceed the upper normal limit of the assay, affected children, including those aged 3 years and younger, should undergo prophylactic thyroidectomy at a specialist surgical centre in order to remove the need for central node dissection at a later stage.

Disclosure

The authors declare no conflict of interest.