Cushing in a Leaf: Endocrine Disruption From a Natural Remedy

Background:

Information regarding the safety of herbal drugs is often not reported. We describe the case of a 65-year-old woman referred to us for a iatrogenic hypercortisolism, who denied any previous steroid consumption. She reported only a chronic application of a phytocosmetic cream, containing ethanol extract of the Cardiospermum halicacabum (CH) plant. Adrenal insufficiency occurred after the cream application was stopped. CH is used in traditional and Western medicine for its documented anti-inflammatory properties. Once the presence of synthetic glucocorticoids was ruled out in the phytocosmetic product, we investigated whether and how its chronic application could have caused the iatrogenic hypercortisolism.

Methods:

Liquid chromatography high-resolution mass spectrometry (LC-HRMS) was performed to exclude the presence of known glucocorticoids in the cream. ELISA assay and Western blot analysis were employed to assess ACTH secretion and the glucocorticoid receptor expression respectively in murine ACTH-secreting pituitary adenoma cells AtT-20/D16v-F2, treated with dexamethasone, CH tincture, and mifepristone alone or in combination. To detect specific interaction of CH extract with the glucocorticoid receptor, we performed a dual-luciferase reporter assay in HEK293 cells.

Results:

In AtT-20/D16v-F2 cells, CH extract showed to significantly reduce basal and CRH-induced ACTH secretion and the glucocorticoid receptor expression, similarly to dexamethasone; these effects were counteracted by mifepristone. In HEK293 cells, dexamethasone significantly induced luciferase activity after 24- and 36-hour treatment and CH tincture only after 36 hours; these effects were antagonized by mifepristone.

Conclusions:

CH extract displays a glucocorticoid-like activity, by means of a direct binding to the glucocorticoid receptor.

Starting from clinical observation,we demonstrated a glucocorticoid-like action of Cardiospermum halicacabum plant extract on pituitary-adrenal axis mediated by binding to the glucocorticoid receptor.

Herbal drugs have been used since the beginning of human history as remedies for health disturbances. Early in the 19th century, the knowledge about the active principles derived from medicinal plants allowed the development of modern pharmacology, leading gradually to the discovery and development of new synthetic drugs, finally causing herbal drugs to be confined to herbal laboratories. Nonetheless, the herbal drug market has rapidly increased in the past few decades, both in the United States and Europe (1–3), along with the expansion of the so-called “complementary and alternative medicine” (4). Part of the success of the therapies based on herbal drugs (or “phytotherapy”) is most likely due to the belief that “natural is safer”.

Nevertheless, in Western medicine, recourse to synthetic glucocorticoids is more and more common (5) because they are used as anti-inflammatory and immunosuppressive agents in several settings. Chronic treatment with glucocorticoids also represents the major known cause of hypercortisolism, a condition more likely to develop during oral therapy, but also after a prolonged topical administration (ie, inhaled, transcutaneous, endo-nasal, intrajoint). Clinical signs of Cushing syndrome caused by chronic exposure to exogenous glucocorticoids typically develop faster than in endogenous hypercortisolism, and are proportionally related to treatment potency and duration (6).

Besides several cases of adulteration of traditional and herbal remedies with synthetic glucocorticoids, to date no case of exogenous hypercortisolism due to herbal drug exposure has been described.

In this study, we describe a case of iatrogenic hypercortisolism during chronic exposure to a topical phytocosmetic product.

The Case

A 65-year-old woman referred to our clinic for typical clinical features of Cushing syndrome (facial plethora, buffalo hump, reddish purple abdominal striae, central obesity). Her clinical history was significant for arterial hypertension, moderate aortic stenosis, previous carotid endarterectomy, type 2 diabetes mellitus, and autoimmune thyroiditis. At presentation, therapy included sartan, potassium canrenoate, bisoprolol, torasemide, acetylsalicylic acid, atorvastatin, and L-thyroxine.

Laboratory tests showed a hypothalamic-pituitary-adrenal (HPA) axis suppression, suggesting iatrogenic hypercortisolism (Table 1). Abdominal computed tomography scan, showing a significant hepatomegaly, was negative for adrenal lesions; pituitary magnetic resonance imaging was normal.

The patient denied any previous steroid use and, carefully questioned about this, she denied the use of intranasal sprays or joint injections containing steroids. However, she reported a daily application of a phytocosmetic cream for the treatment of a submammary erythema for approximately 3 years. The active principles of the cream were: 10% ethanol extract of Cardiospermum halicacabum (CH) plant, 0.5% bisabolol, 0.3% 18β-glycyrrhetinic acid.

In addition, the patient developed clinical and biochemical features of adrenal insufficiency after the cream administration was stopped. An ACTH stimulation test (250 μg) was then performed, showing no response of cortisol and dehydroepiandrosterone sulfate. Replacement therapy with cortisone acetate (25 mg daily) was started, Cushing somatic features progressively resolved and the adrenal function partially recovered after three years (Table 1).

Table 1.

Laboratory Tests

Test	Result
	Basal			Reference Limits
Na	137 mmol/liter			135–145
K	4.3 mmol/liter			3.3–5.1
Serum cortisol	44 nmol/liter			138–690
Urine free cortisol	30 nmol/24 h			90–694
DHEAS	<0.4 μmol/liter			0.9–11.7
	Basal	3 Mo Later	4 Y Later
ACTH	<5 ng/liter	<5 ng/liter	45 ng/liter	10–50
ACTH Test (250 μg)	Basal	3 Mo Later	4 Y Later
Serum cortisol at time 0 min	45 nmol/liter	87 nmol/liter	171 nmol/liter
Serum cortisol at time 60 min	55 nmol/liter	632 nmol/liter	221 nmol/liter

Abbreviation: DHEAS, dehydroepiandrosterone sulfate.

Serum ACTH and ACTH stim test were performed just after, 3 mo, and 4 y after cream discontinuation.

The purpose of this study is to investigate whether and by which mechanism the chronic application of the herbal cream could have caused a clinical and biochemical pattern of iatrogenic hypercortisolism, as in the described clinical case. Cardiospermum halicacabum (family Sapindaceae) is an herbaceous climber widely found in tropical and subtropical Asia and Africa and largely used in traditional medicine for many purposes, particularly for its anti-inflammatory properties (7–11). In Western medicine, it is mainly used as hydro-alcoholic extract (mother tincture) in creams and ointments for the treatment of atopic dermatitis, eczema, and psoriasis (12, 13).

Materials and Methods

Halicalm cream was purchased from Labo Phyto Tre S.r.l. Whole plant mother tincture of CH was purchased from Laboratoires Lehning SAS. HPLC grade acetonitrile, formic acid, betamethasone, methylprednisolone, and budesonide were purchased from Sigma-Aldrich. Purified water from Milli-Q System (Millipore) was used to prepare mobile phase and standard solutions.

Nitrocellulose syringe filters (0.20 μm) were purchased from Phenomenex. Stock standard solutions of betamethasone, methylprednisolone, and budesonide were prepared by dissolving each compound at a concentration of 1 mg/ml in acetonitrile. Stock solutions were stored at 4°C for at most 1 month; further dilutions were prepared daily.

Dexamethasone, mifepristone, and CRH were purchased from Sigma-Aldrich and were diluted in charcoal dextran-stripped culture medium.

Liquid chromatography high-resolution mass spectrometry analysis

The extraction of the compounds of interest from the cream was achieved by following a validated published procedure (14): 0.5 g of Halicalm cream, containing CH ethanol extract, was fortified with the internal standard (methyltestosterone) at the final concentration of 20 μg/g and diluted acetonitrile (2 ml). The mixture was heated up to 45°C, sonicated for 30 minutes, and diluted with water/acetonitrile 50/50 v/v (5 ml). The solution was then filtered through a membrane (0.45 μm). Liquid chromatography mass spectrometry (LC-MS) analysis was carried out on a Surveyor LC system (Thermo) coupled with a high-accuracy, high-resolution Orbitrap XL mass spectrometer (Thermo) equipped with an electro-spray ionization source operating in positive ion mode. Other working parameters were set as follows: transfer capillary 300°C; electro-spray ionization needle spray voltage +4.0 kV; sheath gas 30 au; auxiliary gas 5 au. Mass spectra were acquired at a resolution of 30 000.

LC was performed on a Surveyor binary pump (Thermo) equipped with an Atlantis T3 Waters (150 mm × 2.1 mm × 2.5 μm) kept at 40°C. The following mobile phases were used: A (0.1% formic acid in water), B (0.1% formic acid in acetonitrile) delivered in a linear gradient, t = 0–5 minutes 80% A, t = 20 minutes 50% A, t = 25 minutes 5% A, t = 27 minutes 5% A, t = 30 minutes 80% A. The flow was kept constant at 300 μl/min. The injection volume 20 μl full loop. Analytes were identified by searching the exact mass of the ion [M+H]⁺.

Cell culture, ELISA assay, and Western blot analysis

The murine ACTH-secreting pituitary adenoma cells AtT-20/D16v-F2 were obtained from ATCC (American Type Culture Collection) and cultured in DMEM (Life Technologies) with 1% penicillin/streptomycin (EuroClone) and 10% horse serum (EuroClone) (15).

ACTH “Ultra Sensitive” lumELISA kit (Calbiotech) was used as previously described (16). The sensitivity was less than 1 pg/ml at the 95% confidence limit. Intra- and interassay variation coefficients were 6 and 8.7%, respectively. ACTH levels were assessed in the conditioned culture medium of AtT-20/D16v-F2 cells treated for 48 hours with 1 μM dexamethasone, 2.5% CH tincture, and 1 μM mifepristone alone or in combination. Dexamethasone was used as reference glucocorticoid (GR agonist), while mifepristone was used as reference GR antagonist. Results are expressed as mean value ± standard error percent relative light unit vs control cells from three independent experiments in seven replicates.

Western blot analysis was performed to evaluate GR alpha (GR-α) protein expression in AtT-20/D16v-F2 cells. A total of 30 μg of protein extract for each sample was fractionated on 10% SDS-PAGE and transferred by electrophoresis to nitrocellulose membranes (Protran). Subsequently, the membranes were incubated at 4°C overnight with 1:1000 polyclonal rabbit anti-β-actin and 1:1000 antiglucocorticoid receptor antibody (both from Cell Signaling). Secondary antibodies (1:2000 antirabbit and 1:5000 antimouse horseradish peroxidase-conjugated IgG, both from Dako Italia) were incubated for 1 hour at room temperature and binding was revealed using enhanced chemiluminescence Western blotting detection reagents (Pierce).

Transfection and Dual Glo luciferase assay

The GloResponse 9XGAL4UAS-luc2P HEK293 cell line (Promega) is generated by clonal selection of human embryonic kidney 293 (HEK293) cells, stably transfected with the pGL4.35 [luc2P/9XGAL4UAS/Hygro] vector. This vector is specifically responsive to glucocorticoids. GloResponse 9XGAL4UAS-luc2P HEK293 cells were cultured in DMEM (Life Technologies) supplemented with 1% penicillin/streptomycin, 10% fetal bovine serum (EuroClone), and 200 μg/ml hygromycin B (Sigma-Aldrich). The cells were maintained at 37°C in a humidified atmosphere with 5% CO₂. GloResponse 9XGAL4UAS-luc2P HEK293 cells were seeded in 96-well culture plates at 1 × 10⁴ cells/well, using phenol red-free DMEM (Life Technologies) containing 5% charcoal/dextran-treated fetal bovine serum (EuroClone). The next day the cells were transfected with 100 ng pBIND-GR Vector (Promega) containing the GR ligand-binding domain, using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer's instruction. One day after transfection, the cells were treated with 1 μM dexamethasone, 2.5% CH tincture, and 1 μM mifepristone alone or in combination. Control cells were treated with vehicle solution (1.625% ethanol). After 24 and 36 hours, cells were assessed for firefly and Renilla luciferase activity using the dual-luciferase reporter assay (Promega) according to the manufacturer's instruction. Chemiluminescence was measured by EnVision Multilabel Reader (PerkinElmer). Results are expressed as mean value ± SE percent relative light unit vs control cells from three independent experiments in four replicates.

Statistical analysis

Paired or unpaired Student t test was used to assess individual differences between means. P value <.05 was considered significant.

Results

First, we ruled out the presence of synthetic glucocorticoids in the cream containing CH extract by means of a LC-HRMS method that excluded the presence of the analytes reported in Table 2. The limit of detection of the technique was 1 μg/g of cream.

Table 2.

Exact Mass of the Glucocorticoids Screened in Halicalm Cream

Analyte	Exact Mass MH+	Analyte	Exact Mass MH+
Beclomethasone	521.23006	Fluticasone	445.16549
Betamethasone	393.20718	Fluticasone propionate	501.19171
Budesonide	431.24281	Hydrocortisone	363.21664
Cortisone	361.20095	Meprednisone	373.20095
Cortisone acetate	403.21152	Methylprednisolone	375.21660
Deflazacort	422.22241	Paramethasone	393.20718
Dexamethasone	393.20718	Prednisolone	631.20095
Flunisolide	435.21774	Triamcinolone	395.18644
Fluocortolone	377.21226	Triamcinolone acetonide	435.21774

To verify a possible glucocorticoid-like effect of CH tincture, we tested whether CH tincture influences basal and CRH-stimulated ACTH secretion in AtT-20/D16v-F2 cells.

As shown in Figure 1, basal ACTH levels were significantly reduced in conditioned medium from AtT-20/D16v-F2 cells treated with either dexamethasone (−29%, P < .01) or CH tincture (−23%, P < .01) as compared to control untreated cells. GR antagonist mifepristone did not significantly affect ACTH levels, but significantly reduced the inhibitory effect of dexamethasone and CH tincture on this parameter (+22% vs dexamethasone and +19% vs CH tincture, with P < .01 and P < .05, respectively).

Figure 1.

Effects of CH tincture on basal and CRH-induced ACTH secretion.

AtT-20/D16v-F2 cells were incubated in 96-well plates for 48 hours in culture medium supplemented with 1 μM dexamethasone, 2.5% CH tincture, and 1 μM mifepristone alone or in combination. Control cells were treated with vehicle solution. ACTH levels were determined as described in the “Materials and Methods” section, before (white bars) and after treatment with 100 nM CRH (black bars). **P < .01 vs control; ##P < .01 vs CRH-treated cells.

CRH significantly stimulated ACTH secretion (+125% vs control untreated cells, P < .01). Dexamethasone and CH tincture significantly reduced CRH-induced ACTH secretion (−33% and −49% vs CRH treated cells, respectively, both with P < .01). Mifepristone significantly reduced CRH-induced ACTH secretion (−14% vs CRH-treated cells, P < .01), and significantly reduced the inhibitory effect of dexamethasone and CH tincture on this parameter (+14% vs CRH + dexamethasone and +19% vs CRH + CH tincture, both with P < .01).

Briefly, CH tincture reduces basal and CRH-induced ACTH secretion in AtT-20/D16v-F2 cells supporting the hypothesis that CH tincture may have a glucocorticoid-like effect.

The effects of CH tincture on GR-α expression were then explored. As shown in Figure 2, GR-α levels decreased after treatment of AtT-20/D16v-F2 cells with either dexamethasone or CH tincture as compared to control. The GR antagonist mifepristone slightly reduced GR-α expression and counteracted the inhibitory effect of dexamethasone and CH tincture on this parameter. These results further support the hypothesis that CH tincture may have a glucocorticoid-like effect. Finally, to evaluate whether CH tincture directly binds to GR, we transfected GloResponse 9XGAL4UAS-luc2P HEK293 cells with pBIND-GR vector. Later, the cells were treated with 1 μM dexamethasone, 2.5% CH tincture, and 1 μM mifepristone alone and in combination. After 24 and 36 hours, luciferase activity was measured using dual-luciferase reporter assay. As shown in Figure 3, after 24 hours, dexamethasone significantly induced luciferase activity (+2521% vs control, P < .01) that was not modified by CH tincture. Mifepristone reduced luciferase activity (−64% vs control, P < .01) and counteracted the stimulatory effects of dexamethasone (−1785% vs dexamethasone, P < .01). Moreover, mifepristone reduced luciferase activity also in the presence of CH tincture (−87% vs control, P < .01). After 36 hours, both dexamethasone and CH tincture significantly induced luciferase activity (+9929% and +4829% vs control, both with P < .01). Mifepristone significantly reduced luciferase activity (−51% vs control, P < .05) and counteracted the stimulatory effects of both dexamethasone (−7345% vs. dexamethasone, P < .01) and CH tincture (−4019% vs. CH tincture, P < .01). These results demonstrate that the CH tincture directly and specifically binds GR, supporting its glucocorticoid-like activity.

Figure 2.

Effects of CH tincture on GR-α expression.

AtT-20/D16v-F2 cells were incubated in culture medium for 48 hours in the presence of 1 μM dexamethasone, 2.5% CH tincture, and 1 μM mifepristone alone or in combination. Control cells were treated with vehicle solution. GR-α levels were assessed by Western blot as described in the “Materials and Methods” section. β-actin is shown as a loading control.

Figure 3.

CH tincture directly binds GR-α.

GloResponse 9XGAL4UAS-luc2P HEK293 cells were incubated in 96-well plates for 24 (white bars) or 36 hours (black bars) in culture medium supplemented with 1 μM dexamethasone, 2.5% CH tincture, and 1 μM mifepristone alone or in combination. Control cells were treated with vehicle solution. *P < .05 and **P < .01 vs control.

Discussion

Although the laboratory pattern of our clinical case was evocative of a iatrogenic hypercortisolism, the patient denied any glucocorticoid consumption in the past several months, only reporting chronic application of a phytocosmetic cream containing CH extract. Supported by the development of adrenal insufficiency after cream withdrawal, the following step was to rule out a possible sophistication of the cream with synthetic glucocorticoids. Several reports of herbal remedy adulterations with these drugs have already been described before in some cases of suspected hypercortisolism, also after administration of apparently innocuous and above suspicion products of the so-called “traditional medicine” (17–22). In our case, however, LC-HRMS analysis allowed us to exclude the presence of known synthetic glucocorticoids in the cream, with a reliable limit of detection.

Our research then focused on the components of the phytocosmetic product, particularly on Cardiospermum halicacabum, a plant whose anti-inflammatory properties are well known in Asiatic and African traditional medicine. The extract of this plant is mainly used in Western medicine for the topical treatment of several dermatological pathologies, where it is considered as a “natural cortisone.” Previous studies have confirmed anti-inflammatory, antipyretic, and analgesic properties of CH in animal models (23) and in vitro. Some authors demonstrated that CH ethanol extract suppresses the production of TNF-α and nitric oxide by human peripheral blood mononuclear cells (24). Another study documented that CH ethanol extract in mouse macrophage cell lines dose-dependently inhibits messenger RNA expression of cyclooxygenase-2, inducible nitric oxide synthase, TNF-α, and cyclooxygenase-2 protein expression; these effects could probably be mediated by blocking nuclear factor-κB activation (9). However, to date, a specific glucocorticoid-like action of CH and, in particular, the influence of CH on HPA axis has not been described.

Glucocorticoids exert a renowned negative feedback on the pituitary through the interaction with the GR, which acts as a nuclear transcription factor. In ACTH-secreting cells glucocorticoids bind to GR, leading to the transcriptional repression (so-called “transrepression”) of proopiomelanocortin gene and then to the inhibition of ACTH synthesis (25, 26). In our settings, CH was capable of inhibiting basal and CRH-induced ACTH secretion, in a similar fashion to dexamethasone. These data confirm the hypothesis that CH has a specific glucocorticoid-like activity, because it is antagonized by the GR antagonist mifepristone.

Activation of GR by its cognate ligand is involved in the receptor downregulation in most cell lines and animal tissues. This effect of “homologous down-regulation” of GR seems to take place both at a transcriptional and posttranslational level (eg, GR protein turnover) (27–30), but the question is still open to debate, particularly concerning the different cell- and tissue-specific differences observed in several studies. Our Western blot results show that CH tincture reduces GR-α protein expression in AtT-20/D16v-F2 cells, similarly to the reference glucocorticoid dexamethasone, and furthermore that mifepristone antagonizes this effect for both dexamethasone and CH tincture. In addition, the direct and specific interaction of CH tincture with GR is evident in HEK cells engineered to show the induction of GR-DNA binding by exogenous stimuli. In these settings, DNA binding by GR was induced by dexamethasone after both 24 and 36 hours, whereas CH tincture induced this effect only after 36 hours. Once again, these data confirm a glucocorticoid activity of CH, which, on the other hand, appears to be less potent in comparison to the employed reference glucocorticoid dexamethasone. In addition, the specificity of these effects has been confirmed by the evidence that GR-DNA interaction induced by both dexamethasone and CH are blocked by the GR antagonist mifepristone.

Overall, the data described in this study allow us to sustain that CH extract acts upon the HPA axis with a glucocorticoid-like mechanism by binding to the GR. To date, this is the first description of a iatrogenic hypercortisolism caused by a herbal drug.

However, it is unrealistic to affirm that we are talking about a true “natural cortisone” with the same efficacy of synthetic glucocorticoids. It is very unlikely, indeed, that glucocorticoid effects of CH have not been brought up before, given the widespread and ancient use of this natural remedy in traditional Asian and African medicine and its increasing utilization in Western herbal medicine. Most probably, the inveterate application of the cream by our patient was decisive for the development of hypercortisolism with HPA axis suppression. As the famous physician Paracelso wrote in the fifteenth century: “Dosis sola facit ut venenum non fit” (Only the dose distinguishes a drug from a poison).

Another aspect worthy of consideration is the possibility of a particular individual susceptibility to glucocorticoids in our patient. At present, this aspect represents a growing field of interest in clinical research, because variability in glucocorticoid sensitivity, particularly mediated by GR polymorphisms (31), seems to have implications in many clinical aspects (eg, response to glucocorticoid therapy, different phenotypes in Cushing syndrome and in several chronical diseases).

It is important to emphasize that the patient's adrenal function did not fully recover; we do not know whether the patient's adrenal function was sufficient before she started the cream use, but nothing suggests a preexisting adrenal insufficiency. Even after ACTH normalization, cortisol secretion remained insufficient. We suppose that some components of CH extract could have had a direct toxic effect on the adrenal glands and it might have prevented the full recovery of adrenal function, although we are not able to prove it. This toxic effect could be elicited by chronic application of the cream.

Nonetheless, the clinical case described in this study points out that herbal drugs must not be underestimated concerning their possible dangerous effects, and that it is necessary to improve knowledge about their pharmacodynamic and pharmacokinetic properties. This should become mandatory in a time in which phytotherapy use is growing in Western society without a corresponding progress in pharmacognosy (ie, the branch of pharmacology dealing with herbal drugs). It is worth remembering that research about herbal drugs is made intrinsically difficult for the peculiar characteristics of these remedies, compounds of multiple active principles that are overall responsible for the therapeutic profile. Another source of variability is represented by the different methods of cultivation and collection, and above all the different manufacturing leading to the final plant extract. All of these aspects make it difficult to obtain a standardization of herbal drugs comparable to conventional drugs.

It should also be considered that this issue concerns not only scientific and safety aspects, but also involves legal consequences. Herbal drug marketing is indeed not constrained by the same strict regulation as for conventional drugs, which involves preclinical and clinical trials.

Finally, coming back to our clinical case, it will be interesting to find which element or combination of elements contained in CH tincture could have played a role to determine the glucocorticoid-like effect. However, for the reasons discussed here, it will be a very intriguing challenge.

Acknowledgments

The authors thank Giovanni Giordano, MD, who first evaluated the patient and referred her to our attention.

This work was supported by grants from the Italian Ministry of Education, Research and University (FIRB RBAP11884 M, RBAP1153LS), and Associazione Italiana per la Ricerca sul Cancro in collaboration with Laboratorio in rete del Tecnopolo “Tecnologie delle terapie avanzate” of the University of Ferrara.

Disclosure Summary: C.M., E.Z., M.B., M.d.R., E.G., and A.N. declare no competing interest. E.d.U. reports grants and personal fees from Novartis, grants from Sanofi, and grants from Prostrakan outside the submitted work. M.C.Z. reports personal fees from Novartis and personal fees from Genzyme outside the submitted work. R.V. reports grants and personal fees from Novo Nordisk, personal fees from Sanofi, personal fees from Orexigen, and personal fees from AstraZeneca outside the submitted work. All authors declare no financial relationships with any organization that might have an interest in the submitted work in the previous three years and no other relationships or activities that could appear to have influenced the submitted work.

C.M. and E.Z. contributed equally to this article and should both be considered first authors.

Abbreviations

 
• CH

  Cardiospermum halicacabum

 
• GR

  glucocorticoid receptor

 
• HPA

  hypothalamic-pituitary-adrenal

 
• LC-HRMS

  liquid chromatography high-resolution mass spectrometry.