Berardinelli-Seip Syndrome: Report of an Old Case Successfully Treated with Anti-Glucocorticoid Therapy Followed by Bilateral Adrenalectomy

Abstract

A female teenager was diagnosed in 1986 with Berardinelli-Seip syndrome (congenital generalized lipodystrophy). Following the predictable failure of the usual treatments for her severe type 2 diabetes and hypertriglyceridemia, we decided to treat her with a novel anti-glucocorticoid-centered approach. In 1988, we treated her with mifepristone alone (9 weeks), then with mifepristone combined with ketoconazole (1 week), and again, with mifepristone alone (2 weeks). Acanthosis nigricans, as well as eruptive xanthomas, experienced complete regression following the anti-glucocorticoid therapy. Moreover, the patient gained 7 kilograms. Besides, there was a striking metabolic amelioration with mifepristone therapy. The addition of ketoconazole strongly reduced the relevant mifepristone-induced hypercortisolemia within 1 week. Fasting serum glucose, insulin, and triglycerides fell from day 1 to day 7 without reaching values within the normal range. Two weeks after ketoconazole withdrawal (while keeping mifepristone administration), serum triglyceride and glucose values rose significantly. Eleven days after bilateral adrenalectomy, fasting glucose values were within normal limits or slightly above. An oral glucose tolerance test (75-g OGTT) performed 13 days after surgery showed insulin values within normal limits, fasting serum glucose values within the normal range, and a 2-h serum glucose value in the diabetic range. These findings were consistent with our working hypothesis proposing that Berardinelli-Seip syndrome is due to cortisol-mediated unrestrained lipolysis.

1. Introduction

Congenital Generalized Lipodystrophy (CGL, Berardinelli-Seip syndrome, BSCL) is a rare autosomal recessive disorder with a prevalence in the range of 1–10 patients per million people [1, 2]. However, in northern areas of Brazil, its prevalence is much higher, amounting to 32.3 patients per million people [3]. We recognize at least four CGL types: Type 1, due to mutations in the AGPAT2 gene, which is located on chromosome 9q34 and encodes the enzyme 1-acylglycerol-3-phosphate O-acyltransferase 2 (AGPAT2); Type 2, due to mutations in the BSCL2 gene, which is located on chromosome 11q13 and encodes Seipin; Type 3, due to mutations in the CAV1 gene, which is located on chromosome 7q31 and encodes caveolin-1; and Type 4, due to mutations in the polymerase I and transcript-release factor [PTRF] gene, which is located on chromosome 17q21.2 and encodes cavin [1].

About 95% of CGL patients harbor either AGPAT2 or seipin mutations. The most common CGL is Type 1, and the rarest is Type 3. In Brazil, over 90% of CGL cases are CGL type 2, while in Europe, Middle East, and Japan, most patients have CGL Type 1. CGL patients have an extreme deficiency of fat depots. Therefore, most CGL patients have low serum leptin and adiponectin levels. They usually have elevated circulating serum triglycerides (very-low-density lipoproteins plus chylomicrons) and develop severe steatosis within their lean organs, such as the liver and skeletal muscles. Lipotoxicity of these organs produces in CGL patients an insulin-resistant state that does not respond to current treatments.

Two key metabolic features are prominent in these patients: much-elevated gluconeogenesis along with much-reduced insulin-mediated skeletal muscle glucose uptake. Usually, by the second decade of life, these patients develop type 2 diabetes requiring very high insulin doses. Hypertriglyceridemia in these patients is very difficult to treat, and it may manifest itself as cutaneous xanthomas. The best current treatment for CGL is recombinant leptin [4], but this is not widely available. Besides, leptin therapy is associated with weight loss in these patients.

The first two reported patients—both of pediatric age—were described in Brazil by Dr. Waldemar Berardinelli in 1954 [5]. In 1959, the Norwegian pediatrician Martin Seip published the report of three additional patients [6]. Berardinelli-Seip syndrome has been hard to understand and therefore, very difficult to treat.

To advance the knowledge on CGL, several groups attempted to generate transgenic murine models of the disease. In 1998, two transgenic models of murine lipoatrophy were simultaneously published [7, 8]. The A-ZIP/F1 transgenic mouse model had an extreme fat deficiency, whereas the aP2-SREBP-1c mouse model had only a moderate fat deficiency.

In 1994, leptin (a cytokine produced by the adipocytes having endocrine actions) was discovered [9]. In 1999, Shimomura et al. [10] showed that recombinant leptin reversed diabetes in the aP2-SREBP-1c mouse model of CGL and concluded that insulin resistance was secondary to the severe hypoleptinemia found in these animals. However, the A-ZIP/F1 mouse model responded poorly to exogenous leptin [11]. By contrast, in 2002, a study demonstrated that total adrenalectomy in A-ZIP/F1 diabetic mice induced a substantial metabolic improvement by increasing liver and muscle insulin sensitivity [12]. The authors attributed these improvements to the adrenalectomy-induced disappearance of chronic hypercorticosteronemia. Moreover, leptin treatment in A-ZIP/F1 mice reduced their elevated circulating corticosterone levels. So, it was conceivable that hypoleptinemia was behind the adrenal axis overactivation and the subsequent hypercorticosteronemic state exhibited by this mouse model.

In 2002, recombinant methionyl human leptin (metreleptin) reversed insulin resistance in nine women with congenital or acquired lipodystrophy and serum leptin levels <4 ng/mL [4]. In 2014, recombinant human leptin (metreleptin; Myalept®) received FDA approval to treat lipodystrophies [13]. In 2009, Cortés et al. generated the AGPAT2-deficient mouse, a transgenic animal model of lipoatrophy resembling CGL Type 1 [14]. In 2011, Cui et al. [15] reported that seipin ablation in mice results in severe generalized lipodystrophy. In 2012, Chen et al. [16] confirmed that inactivation of seipin in mice leads to severe lipodystrophy. Also, these authors shed light on the mechanisms involved in the process. They found that in vitro differentiation of murine embryonic fibroblast and stromal vascular cells had normal early-phase adipocyte differentiation, but a striking failure of terminal differentiation. This unsuccessful adipogenesis was secondary to a runaway cyclic AMP-dependent lipolysis and silencing of the transcription factors regulating adipogenesis. In vitro adipogenesis was rescued by inhibitors of lipolysis, but not by peroxisome proliferator-activated receptor (PPAR)-gamma agonists, such as pioglitazone. A recent review [17] suggests a central role of unrestrained lipolysis in the genesis of lipoatrophy of seipin-deficient individuals. In summary, seipin stimulates adipogenesis and inhibits cyclic AMP-dependent lipolysis.

The pathophysiology of the AGPAT2-deficient patients is cloudier compared with the situation of the homologous seipin-deficient patients. In 2016, Cautivo et al. [18] showed that the AGPAT2 gene is essential for the postnatal development and maintenance of white and brown adipose tissue.

A simplistic belief is that AGPAT2 deficiency impairs lipogenesis, while seipin deficiency impairs normal adipogenesis. Both conditions result in triglyceride (TG)-depleted adipocytes. However, a TG-depleted adipocyte also results from ablation of perilipin in murine adipose tissue. Nevertheless, in the latter situation, the TG-depleted adipocyte secretes an increased, rather than a reduced, amount of leptin [19]. Thus, a TG-depleted adipocyte is not necessarily associated with hypoleptinemia. In other words, the exact mechanism by which AGPAT2 deficiency leads to hypoleptinemia is unknown. Overall, hypoleptinemia seems to be a commonality in generalized lipodystrophies. Therefore, the biggest investigational challenge is to figure out how hypoleptinemia and severe insulin resistance are linked together in CGL.

Herein, we report an extraordinary experience with a patient with Berardinelli-Seip syndrome (1986–1988) seen before the leptin era. At that time, we hypothesized that CGL was somehow the consequence of the local excess of cortisol action on the adipocyte. To test our daring hypothesis, we used mifepristone, a potent anti-glucocorticoid drug. Having previous experience with the drug on a patient with a previously operated-on, recurrent ectopic adrenal cancer and severe Cushing’s syndrome [20], we anticipated that mifepristone would probably produce an overactivation of the hypothalamic–pituitary–adrenal axis. For this reason, following 9 weeks of mifepristone therapy alone, we briefly added ketoconazole to the treatment to partially block cortisol synthesis. We devised this therapeutic strategy to reduce serum cortisol levels, seeking to reinforce the anti-glucocorticoid effect of mifepristone. Finally, we stopped ketoconazole to reduce the anti-glucocorticoid action of the combined intervention. Overall, our results with the abovementioned anti-glucocorticoid approach permitted us to surmise that total adrenalectomy would benefit the patient.

1.1 Case report

A 16-year-old female patient entered the Endocrine Unit of the University Hospital (Hospital José Joaquín Aguirre, Universidad de Chile, Santiago) with a recent diagnosis of type 2 diabetes. Her parents were first cousins. Since she was born, pediatricians were intrigued by her peculiar appearance, characterized by scarcity of subcutaneous fat, muscular prominence, and abdominal distension.

Our patient had an acromegaloid face, scarcity of subcutaneous adipose tissue, conservation of mechanical fat, severe acanthosis nigricans, prominent veins, and muscular prominence. She had a voracious appetite and exhibited eruptive xanthomas (sparing the face and the chest) especially over her palms and elbows.

She also had clinical hyperandrogenism, including facial and scalp seborrhea and mild clitoromegaly. She had a history of recurrent periods of amenorrhea. Her liver and spleen were notoriously enlarged, producing a prominent abdomen. A mild thyromegaly and a small umbilical hernia were present. The normal intellectual development and the presence of mechanical fat (located in palms, soles, joints, and retro-orbital space) in our patient suggested that she was AGPAT2-deficient rather than seipin-deficient. We did not have her DNA sequenced since this technique debuted in CGL cases just in the current century.

Fasting serum glucose, insulin, and triglycerides were 225 mg/dL, >400 mU/L, and 7400 mg/dL, respectively (normal levels: ≤99 mg/dL, ≤13 mU/L, and ≤ 150 mg/dL, respectively). The patient was diagnosed with Berardinelli-Seip syndrome and polycystic ovary syndrome.

We treated her with high insulin doses with disappointing results, which attested to the presence of an extreme insulin resistance. Given the acute pancreatitis risk, we carried out an unsuccessful attempt to reduce her extremely high levels of serum triglycerides with high doses of omega-3-rich fish oil (up to 20 grams per day). Our medical staff thoroughly discussed the case and reached a consensus: she was not treatable with conventional medications.

1.2 Rethinking the patient from scratch

We realized that, to help the patient, we had to rethink her. For us, the extreme scarcity of body fat in our patient was the key to understanding how to deal with her disease. We reviewed the available literature (years 1986–1988) regarding adipogenesis, lipogenesis, and lipolysis. We looked at the fat storage within the adipocyte like a “bank account” of fat. An “empty adipocyte” could result from deficient lipogenesis or increased lipolysis.

We learned that serum insulin levels stimulate lipoprotein lipase (LPL). LPL provokes the lipolysis of circulating serum chylomicrons and very-low-density lipoprotein particles (VLDL). Then, free fatty acid (FFA) molecules enter the adipocyte to initiate lipogenesis. Three FFA molecules bind to a single acyl-glycerol molecule to form the triacylglycerol molecule (triglyceride or TG) stored within lipid droplets inside the cytoplasm. The stored TG molecule may be subjected to a hormone-sensitive lipase (HSL)-mediated lipolysis, releasing FFAs and glycerol into the circulation. Even minute amounts of circulating insulin can tonically inhibit HSL.

When hepatic glycogenolysis is exhausted at dawn, serum glucose and insulin levels fall, interrupting the HSL inhibition. For us, a piece of crucial information was that cortisol exerts a permissive role on the HSL activation in the cytosol of the adipocyte. In other words, the HSL is inactive in the absence of cortisol inside the cytosol of the adipocyte. Cortisol reaches the cytosol of the adipocyte through the internalization of extracellular, inactive cortisone. This inactive cortisone is transformed inside the cytosol into physiologically active cortisol by the enzyme 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1). In turn, cortisol stimulates HSL acutely, whereas it stimulates LPL chronically. Acute lipolysis occurs when energy is required: the released FFA and glycerol molecules provide the liver with substrates to increase glucose production at dawn. These tips summarize what we learned about lipogenesis and lipolysis by reviewing the available literature. By understanding the physiological roles of leptin, our comprehension of adipose tissue physiology has grown a great deal. This new knowledge will allow us to reinterpret our extraordinary findings.

1.3 The decision to initiate an anti-glucocorticoid-centered, experimental therapeutic approach

Theoretically, fat depletion in CGL may be secondary to defects either in adipogenesis or lipogenesis. However, other possibilities may exist. We hypothesized that cortisol-mediated, unrestrained lipolysis was at the core of the CGL in our patient.

The main advantage of our daring—and even naïve—hypothesis was its testability. Serendipitously, we did have access to mifepristone (RU-486), a potent anti-progesterone and anti-glucocorticoid steroidal drug. We had previously treated with mifepristone for 5 months a patient with recurrent hepatic adrenal rest cancer-mediated hypercortisolism. Before that experience, back in 1985, we had postulated in a Lancet letter [21] that glucocorticoid-producing adrenal cancers might be glucocorticoid-dependent. We had observed a striking, rapid disappearance of liver and lung metastases after using ketoconazole (1200 mg/day) on a female patient to block excessive cortisol synthesis by adrenal cancer. Roussel-Uclaf kindly donated 1000 pills of mifepristone (200 mg each) to treat this patient with an ectopic adrenal cancer. That trial taught us what to expect from the mifepristone administration. However, when the hypercortisolism spectacularly faded away after 5 months of treatment, the patient declared herself cured, refusing further treatment. Unfortunately, she died less than a year after stopping treatment. We, therefore, were left with a substantial amount of available mifepristone pills.

Also, we had extensive experience with ketoconazole in several cases of Cushing’s syndrome (seen from 1983 through 1988). In 1983, we successfully used ketoconazole to treat hypercortisolism in a patient with an adrenal rest tumor of the liver. This success permitted us to remove her ectopic adrenal tumor from the liver, resulting in an apparent surgical cure. We reported this experience in 1985 [20]. Up until then, nobody else had published a clinical trial with ketoconazole in Cushing’s syndrome. This initial, positive experience paved the way for us to acquire expertise in using ketoconazole in Cushing’s syndrome.

The endocrinologists of our Unit discussed the complex situation of our patient with CGL. At the time, we did not have an Ethics Committee at the University Hospital. Therefore, our group reached a medical consensus: to offer an experimental mifepristone treatment to the patient and her family. We explained to them that we had nothing else to offer. The patient and her family agreed to receive a trial of mifepristone therapy. We used the same dose of mifepristone (600 mg daily, divided into three 200 mg pills) that we had previously used in our patient with ectopic adrenal cancer. So, after a whole year of unsuccessful therapies, we were ready to proceed with an experimental, unheard-of, therapeutic approach on our 18-year-old patient.

2. Material and methods

2.1 Experimental protocol (1988)

2.1.1 First course of anti-glucocorticoid treatment alone

Mifepristone therapy alone (600 mg/day, given as three 200-mg pills, at 7 AM, 3 PM, and 11 PM) was administered orally for 9 weeks. The anti-glucocorticoid action of the drug would negate the cortisol-mediated negative feedback on secretion of CRF (corticotropin releasing factor) and ACTH (adrenocorticotropic hormone). Therefore, we anticipated an overactive hypothalamic–pituitary–adrenal axis (“adrenal axis”, for short). High circulating cortisol levels will not produce glucocorticoid actions in the presence of mifepristone. Instead, they will stimulate the mineralocorticoid receptor in the distal renal tubule (cortisol-induced hypermineralocortisolism).

For this reason, we measured the mean daily serum cortisol values (individual values were measured at 8 AM, 3 PM, and 11 PM) and urinary free cortisol on day 63 of the trial at the end of this first phase. To monitor the expected metabolic changes, we serially measured serum fasting glucose, insulin, and triglycerides during this period.

2.1.2 Anti-glucocorticoid treatment plus partial blockade of cortisol synthesis

Mifepristone (600 mg/day) plus ketoconazole (800 mg/day, divided into four 200-mg pills given every 6 hours) combination therapy was administered for 1 week.

Ketoconazole was expected to produce a partial blockade of the enhanced mifepristone-induced adrenal cortisol synthesis, thus reducing the prevailing hypercortisolemia. We devised this therapeutic addition to reinforce the anti-glucocorticoid effect of mifepristone. We measured the daily mean serum cortisol values and urinary free cortisol on day 70 of the trial, at the end of the second phase of treatment. We also measured daily serum fasting glucose, insulin, and triglycerides to evaluate the response to the addition of ketoconazole.

2.1.3 Second course of anti-glucocorticoid treatment alone

Mifepristone therapy alone (600 mg/day, given as 200-mg pills, at 7 AM, 3 PM, and 11 PM) was administered orally for 2 weeks. We expected to witness a deterioration of both circulating fasting glucose and triglycerides by stopping ketoconazole. We did not measure cortisol values in serum and urine during this phase.

The whole anti-glucocorticoid intervention (mifepristone alone or combined with ketoconazole) lasted 12 weeks. Acanthosis nigricans and eruptive xanthomas virtually disappeared during the trial (Figure 1). Moreover, the patient gained 7 kilograms.

Figure 1.

Local acanthosis nigricans and eruptive xanthomas, as seen before (A) and after (B) the anti-glucocorticoid intervention. The peculiar aspect of a fat-devoid mesentery at surgery is also shown (C).

3. Results

3.1 Mifepristone therapy alone, 600 mg/day (9 weeks)

This intervention produced a striking amelioration of fasting serum insulin and triglyceride levels, as shown in Table 1. However, as we predicted, the patient experienced an overactivation of the adrenal axis, reflected by very high daily mean serum cortisol and urinary free cortisol excretion at the end of the 9th week (Table 2). The mifepristone-induced adrenal axis overactivity revealed itself as a clinical hypermineralocortisolism. The patient had arterial hypertension (160/100 mmHg), hypokalemia (3.4 mEq/L), and inappropriate urinary potassium loss (56 mEq/day). We interpreted this phenomenon as a mifepristone-induced hypercortisolism overwhelming the cortisol-inactivating capacity of the 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2); this enzyme—located in the distal tubule of the nephron—converts active cortisol into inactive cortisone, thus preventing cortisol-induced activation of the local mineralocorticoid receptor. Mifepristone blocks the glucocorticoid receptor, but not the mineralocorticoid receptor. So, excess cortisol stimulates the mineralocorticoid receptor of the distal kidney tubule, provoking excess renal sodium reabsorption and excess urinary potassium excretion.

Weeks	Serum fasting triglycerides; (mg/dL)	Fasting glycemia; (mg/dL)	Fasting insulinemia; (mU/L)
Baseline	7400	225	>400
Week 1	6310	145	>400
Week 2	3625	250	10.0
Week 3	1220	280	20.0
Week 4	1210	230	18.0
Week 5	1200	290	n/a
Week 6	1500	230	10.0
Week 7	1750	225	11.0
Week 8	1500	224	10.5
Week 9	617	227	8.0

Table 1.

Assessment of metabolic parameters at baseline and after initiation of mifepristone therapy alone (600 mg/day). Mifepristone alone (600 mg/day) reduced serum triglycerides first (1 week) and then reduced serum fasting insulin levels in 2 weeks. Serum fasting glucose levels were unchanged for 9 weeks. At the end of the 9th week, serum levels of triglycerides were just 8.3% of the baseline value. Similarly, fasting insulin levels were less than 2% of the baseline value. Abbreviations: n/a, not available.

Intervention	24-hour urinary-free cortisol (UFC); (mcg/24 h; normal values, ≤100)	Mean daily serum cortisol*; (mcg/dL)
Mifepristone therapy alone; (63rd day)	1130	48.5
Mifepristone plus ketoconazole combination therapy (70th day)	630	26

Table 2.

Assessment of adrenal function at the end of the prolonged mifepristone therapy alone and at the end of the mifepristone plus ketoconazole combination therapy. At the end of the administration of mifepristone alone, 24-h urinary free cortisol (UFC) and mean daily serum cortisol values were grossly elevated, reflecting a mifepristone-induced adrenal hyperactivity. One week after adding ketoconazole to mifepristone, there was almost a 50% decrease in these values, reflecting the ability of ketoconazole to block adrenal cortisol synthesis. *we calculated the mean daily serum cortisol values (individual values were measured at 8 AM, 3 PM, and 11 PM).

3.2 Mifepristone (600 mg/day) plus ketoconazole (800 mg/day) combination therapy (a single week)

As predicted, the addition of ketoconazole (an antifungal agent capable of inhibiting cortisol synthesis) enhanced the anti-glucocorticoid effect of mifepristone, as reflected by a further reduction of the serum levels of triglycerides and fasting insulin and glucose levels (Table 3). These effects paralleled a drop of mean daily total serum cortisol and urinary free cortisol excretion, as shown in Table 2. Table 4 shows a striking reduction of serum insulin values observed during an oral glucose tolerance test (OGTT) performed with 75 g of glucose on the last day of the combined administration, compared with the values observed 2 years earlier.

Days; (Starting from Week 10)	Serum fasting triglycerides; (mg/dL)	Fasting glycemia; (mg/dL)	Fasting insulinemia; (mU/L)
Day 1	380	180	12.0
Day 2	375	145	17.5
Day 3	300	122	13.0
Day 4	250	121	13.5
Day 5	240	135	13.6
Day 6	220	129	7.5
Day 7	230	138	7.0

Table 3.

Assessment of metabolic parameters during mifepristone (600 mg/day) plus ketoconazole (800 mg/day) combination therapy. The addition of ketoconazole to mifepristone for 7 days notoriously reduced serum fasting levels of triglycerides, glucose, and insulin to near-normal values.

Minutes	0’	30’	60’	90’	120’
Basal (1986)	61	193	357	153	>300
Combined Anti-glucocorticoid Administration (1988)	11	19	27	24	23

Table 4.

Serum insulin levels (mU/L) measured during oral glucose tolerance tests (OGTT) performed (with 75 g of glucose) at baseline (1986) and during the mifepristone plus ketoconazole combination therapy (1988). Serum insulin levels observed during the OGTT performed at the end of the combined anti-glucocorticoid administration (1988) were strikingly reduced, as compared with values observed during the baseline OGTT (performed in 1986).

3.3 Second course of mifepristone therapy alone, 600 mg/day (2 weeks)

We observed a progressive deterioration of serum triglyceride and glucose values after stopping ketoconazole (Table 5). Unfortunately, we did not measure cortisol values during this period.

Weeks	Serum fasting triglycerides; (mg/dL)	Fasting glycemia; (mg/dL)	Fasting insulinemia; (mU/L)
11	360	160	n/a
12	438	200	n/a

Table 5.

Assessment of metabolic parameters during the second course of mifepristone therapy alone (600 mg/day). There was a substantial increase in fasting triglyceride and glucose levels after the discontinuation of ketoconazole for 7 days. Abbreviations: n/a, not available.

3.4 Summary of metabolic changes observed during the anti-glucocorticoid intervention

As observed in Table 1, mifepristone therapy alone took 2 weeks to substantially reduce serum fasting triglyceride and insulin levels, while serum fasting glucose was unchanged during this phase of anti-glucocorticoid intervention. The 1-week addition of ketoconazole produced a further reduction of serum triglycerides and a rapid drop of serum fasting glucose values (Table 3). However, over the next 2 weeks of the trial with mifepristone therapy alone, we observed a clear rise in serum triglyceride and glucose values (Table 5).

3.5 The decision to perform total (bilateral) adrenalectomy and its results

None of the observations that we made during our protocol execution negated our working hypothesis. Regarding definitive therapy for lipoatrophic diabetes, mifepristone administration was ruled out for its limited availability and expected complications. By contrast, bilateral adrenalectomy, along with a limited cortisol replacement therapy, would reduce fat exposure to high, lipolysis-inducing cortisol levels. The family and the patient agreed to the surgical procedure. After surgery, we administered a reduced but safe amount of hydrocortisone (15 mg/day, divided into daily doses: 10 mg at 8 AM and 5 mg at 3 PM). In addition, we administered fludrocortisone 0.1 mg daily to avoid excessive urinary sodium loss.

At the time of adrenalectomy, surgeons were surprised to observe a fat-devoid mesentery (Figure 1). The patient recovered uneventfully from the surgical procedure. After adrenalectomy, we measured fasting glucose levels daily, and we performed an OGTT (with 75 g of glucose) 2 weeks later (Table 6).

Minutes	0’	30’	60’	90’	120’
Serum glucose; (mg/dL)	98	140	148	195	210
Serum insulin; (mU/L)	10	45	53	47	52

Table 6.

Serum glucose and insulin values during the oral glucose tolerance test (OGTT) performed (with 75 g of glucose) 13 days following bilateral adrenalectomy, under hydrocortisone (15 mg/day) plus fludrocortisone (0.1 mg/day) replacement therapy. Serum glucose values were within the lowest diabetic range, while hyperinsulinemia was absent.

Twenty-four hours after the adrenalectomy, fasting serum glucose was within the normal range, followed by occasional minimal increments above the normal range in the next 2 weeks. The OGTT results showed a fasting glucose level within the normal range and a 2-h glucose level in the low diabetic range. The insulin values during the OGTT were within the normal range, although with an unusual trajectory (Table 6).

We performed successful adrenalectomy in our lipoatrophic patient 14 years before the adrenalectomy success observed in transgenic A-ZIP/F1 lipoatrophic mice [12].

3.6 Patient discharge and follow-up

We discharged the patient from the hospital (1988) medicated with hydrocortisone (10 mg at 8 AM and 5 mg at 3 PM) plus fludrocortisone (0.1 mg at 8 AM). We instructed her to double the fludrocortisone dose in the summer. Subsequently, she opted to be taken care of by the National Health Service (NHS) due to financial reasons. We never had the opportunity to discuss the case with her new attending diabetologists.

Likely, the idea of using a reduced but safe amount of oral hydrocortisone in this patient did not appeal to her new attending physicians. Besides, the NHS in Chile did not provide fludrocortisone at that time, so the patient likely overdosed with hydrocortisone. Despite these shortcomings, the patient got pregnant twice, having a spontaneous abortion in 2002. In 2003, aged 34, she got pregnant again and delivered a premature, 28-week baby. Her attending obstetricians reviewed the literature and discovered a single case with successful pregnancy in women with CGL [22]. They described the difficult pregnancy in a local obstetrics journal [23]. Even though triglyceride levels remained elevated (below 2000 mg/dL), these levels were substantially lower compared with those observed when we first met her (7100 mg/dL). However, the obstetrics report mentioned one episode of acute pancreatitis antedating her second pregnancy.

We presented our experience in 1989 at The Endocrine Society Meeting in Seattle [24] (Figure 2). However, at that time, we could not write up a paper reporting our findings simply because we could not offer a rational explanation for them.

Figure 2.

The facsimile of our abstract, as it was published in the proceedings of the Endocrine Society Seventy-First Annual Meeting; June 21–24, 1989, Seattle, Washington (USA).

4. Discussion

To reinterpret the therapeutic findings in this teenager, we must briefly revise crucial new knowledge accumulated in the decades since we treated her.

4.1 Extreme adipose tissue insulin resistance (Adipo-IR) at the core of CGL?

Insulin action on the adipocyte stimulates adipogenesis and lipogenesis while inhibiting lipolysis. Extreme insulin resistance in the adipose tissue (Adipo-IR) should severely reduce adipogenesis and lipogenesis. At the same time, Adipo-IR should increase lipolysis, leading to an “empty adipocyte” syndrome. It is conceivable that glucocorticoid action on the adipocyte somehow mediates Adipo-IR. Once the adipocyte becomes triglyceride-depleted, leptin secretion should severely fall, resulting in hypoleptinemia. The latter may lead to adrenal hyperactivity (as seen in murine models of CGL), resulting in hypercorticosteronemia [12]. In patients with CGL, the hypothetical high circulating free cortisol levels (elevated serum total cortisol levels plus hyperinsulinemia-induced low transcortin levels) would perpetuate the increased lipolysis.

Adipo-IR is present at both extremes of adipocyte’s triglyceride storage: triglyceride-replete adipocytes and empty adipocytes (as seen in obesity and CGL, respectively). In both cases, ectopic triglyceride storage in lean organs (liver, muscle, pancreatic beta cell, and skin) replaces further adipocyte triglyceride storage. Lipotoxicity in the liver increases the hepatic glucose output. Lipotoxicity in muscles reduces their insulin-mediated glucose uptake. Finally, beta-cell lipotoxicity induces cell apoptosis. The degree of lean tissue lipotoxicity is severe in CGL patients with “empty adipocytes.” By contrast, in obese subjects, lipotoxicity is considerably lighter (triglyceride-replete adipocytes). If mifepristone administration ameliorates Adipo-IR, it is logical to expect a metabolic improvement of both CGL and obesity.

4.2 Mifepristone and adipose tissue insulin sensitivity

We know that mifepristone improves glucose tolerance and insulin sensitivity in patients with Cushing’s syndrome [25]. However, the specific effect of the drug on adipose tissue’s insulin sensitivity had remained unexplored. In 2021, an NIH group reported that mifepristone improves adipose tissue insulin sensitivity in insulin-resistant individuals [26]. Sixteen overweight or obese subjects with prediabetes or mild type 2 diabetes (without Cushing’s syndrome) received either mifepristone (200 mg/day; 50 mg 4 times a day) or placebo for 9 days with a washout period of 8 weeks. At baseline and following mifepristone and placebo administration, the subjects had a 75-g OGTT and a frequently sampled intravenous glucose tolerance test (FSIVGTT). Whole-body insulin sensitivity was estimated on these subjects calculating three indices: Insulin Sensitivity Index (SI), Matsuda index (MI), and Oral Glucose Insulin Sensitivity Index (OGIS). These indices were not modified by mifepristone 200 mg daily. However, there were significant improvements in the adipose tissue insulin resistance index (Adipo-IR index) (a surrogate marker of fasting adipose-tissue insulin resistance, calculated as the product of fasting insulin and fasting free fatty acids) and in the adipose tissue insulin sensitivity index (Adipo-S_I index, defined as the ratio of the slope of the linear decrease in natural log transformed free fatty acids during the first 90 minutes of the FSIVGTT and the area under the curve of serum insulin during that 90-minute period) [26]. In addition, mifepristone increased insulin clearance but did not modify either insulin secretion or beta-cell glucose sensitivity. Mifepristone use reduced fasting serum glucose, insulin, and triglycerides. Also, the areas under the curve of daily serum ACTH and cortisol values were significantly higher during mifepristone administration. Urinary free cortisol values also rose significantly. Thus, mifepristone 200 mg/day (divided into four 50-mg daily doses) administered to 16 insulin-resistant subjects reproduced our findings using 600 mg daily (divided into three 200 mg doses): serum and urinary cortisol values rose, while serum glucose, insulin, and triglyceride values fell.

Therefore, mifepristone (600 mg/day) should have improved the adipose tissue insulin sensitivity in our patient. Since we did not measure serum FFAs, we cannot evaluate the Adipo-IR index in our patient. However, using a raw estimation of adipose tissue’s insulin resistance—the product of insulin (mU/L) times fasting triglycerides (mg/dL), we obtain the following results: mifepristone alone: at baseline, >2,960,000; 9th week, 4936; mifepristone plus ketoconazole combination therapy: first day, 4560; 7th day, 1610. As a reference, a person with a fasting serum insulin value of 13 mU/mL and 150 mg/dL of fasting triglycerides would have a calculated value of 1950 for this parameter. Therefore, this raw estimation of the adipose tissue’s insulin resistance shows a strikingly positive effect of mifepristone, which is further reinforced by the addition of ketoconazole (>99.9% reduction).

A recent Chinese study on the Adipo-IR index and metabolic syndrome [27] reported mean data from six groups of subjects (three groups for each sex) on insulin and triglyceride values. These data allowed us the calculation of our raw adipose tissue insulin resistance index in 20 control females (612.4), 26 obese women without metabolic syndrome (1165.0), and 85 obese women with metabolic syndrome (3017.9). There was a strong, positive correlation of 0.958 between their published averaged Adipo-IR indices and their corresponding calculated raw Adipo-IR indices.

So, this proposed raw estimation of adipose tissue insulin resistance should be helpful in the clinical setting. Both a serum fasting triglyceride value >130 mg/dL and a serum fasting insulin value >13.2 mU/L suggest the presence of insulin resistance. Therefore, a value above 1716 (13.2 times 130) for this raw index would strongly indicate the presence of adipose tissue insulin resistance. Likely, values for this surrogate index in non-insulin-resistant individuals should be less than 1000 (corresponding to a serum insulin value around 7.5 mU/L and a serum triglyceride value around 130 mg/dL).

4.3 Adrenalectomy and insulin resistance

When we received this patient, we knew that patients with Addison’s disease were lean, insulin-sensitive, and prone to hypoglycemia. On the contrary, patients with Cushing’s syndrome are obese, insulin-resistant, and prone to hyperglycemia. Glucocorticoids promote both obesity and insulin resistance, thus deteriorating diabetes control. By contrast, adrenalectomy benefits diabetes control in patients with Cushing’s syndrome. Of course, we did not know in 1988 that adrenalectomy would improve diabetes in A-ZIP/F1 lipoatrophic mice [12]. Adrenalectomy in this murine model of CGL ameliorates liver and muscle insulin sensitivity. The fact that mice lacking leptin synthesis or leptin action (ob/ob and db/db mice, respectively) are both obese and insulin-resistant [28] suggests that leptin action somehow protects them from insulin resistance. The beneficial effects of adrenalectomy in rodents lacking leptin action [29] indicate that the adrenal gland is necessary for these mice to develop insulin resistance. In summary, we need to find the intermediate steps between insufficient leptin action and insulin resistance. In experimental animals with hypoleptinemia, the adrenal gland appears to mediate the development of insulin resistance.

4.4 The new knowledge on Berardinelli-Seip syndrome and its relationship with leptin

The clinical and metabolic improvements observed in our patient during the anti-glucocorticoid intervention were beyond our expectations. The significant weight gain of our patient is encouraging and particularly intriguing. It suggests de novo storage of triglycerides in the patient’s fat depots. If this latter supposition is true, then one wonders whether the supposedly much-reduced leptin and adiponectin levels at baseline rose when fat progressively accumulated within the adipocytes during anti-glucocorticoid therapy. Future research will probably answer these intriguing questions. Although we realized that the anti-glucocorticoid therapy notoriously improved the abnormal adipocyte physiology of our patient, we did not disclose the mechanism(s) involved. In any case, the beneficial effects of the anti-glucocorticoid treatment support the notion of a detrimental action of endogenous cortisol on the adipocyte physiology in this CGL patient.

Now, we can attempt to offer a rational explanation concerning the effects of mifepristone in our patient with Berardinelli-Seip Syndrome. Nowadays, we know several key pieces of crucial importance, such as the discovery of leptin and its functions [9]. A key concept is that leptin exerts an inhibitory effect on the hypothalamic–pituitary–adrenal axis [30]. Moreover, the loss of leptin-induced inhibition of the hypothalamic–pituitary–adrenal axis provokes gross metabolic dysfunctions. For instance, hepatic gluconeogenesis is largely augmented in murine models of poorly controlled type 1 diabetes, having severely low insulin levels [31]. Shortly after leptin discovery, it was evident that CGL patients exhibited severe hypoleptinemia [32]. In 1998, the new era of transgenic mice with lipoatrophic diabetes introduced revolutionary concepts in the field [7, 8]. The less fat-deficient mice responded well to leptin treatment, whereas the severely fat-devoid A-ZIP/F1 mice responded poorly. These mice exhibit hypercorticosteronemia, indicating an overactivity of their adrenal axes. Intravenous leptin infusions reduce adrenal gland overactivity in these mice. When these A-ZIP/F1 mice were adrenalectomized to abate their hypercorticosteronemia, they experienced significant increments in peripheral and hepatic insulin sensitivity [12]. Again, another piece of evidence is that an adrenal gland is necessary for these leptin-deficient mice to develop diabetes.

Patients with CGL treated with recombinant leptin usually respond well to the hormone [4]. To our knowledge, nobody has yet reported that patients with CGL have an overactive adrenal axis, in parallel with the findings observed in murine models of CGL. Unfortunately, it did not occur to us to evaluate the hypothalamic–pituitary–adrenal axis before the trial began. On the other hand, transcortin (also known as CBG or corticosteroid-binding globulin; the protein produced in the liver that transports cortisol in the blood) expression is reduced by hyperinsulinemia [33]. Patients with CGL should theoretically exhibit high levels of serum free cortisol secondary to both hypoleptinemia-induced adrenal gland overactivity and low transcortin levels (due to hyperinsulinemia-induced reduction in hepatic secretion of transcortin).

If leptin deficiency in humans indeed results in the lack of leptin-mediated inhibition of the hypothalamic–pituitary–adrenal axis, a logical consequence of this hormonal deficiency would be the overactivity of the adrenal glands. This situation would be revealed either by high serum total or free cortisol levels, paralleling the hypercorticosteronemia of leptin-deficient rodents. Lipolysis in insulin-resistant subjects should increase due to resistance to insulin-induced HSL inhibition. If the adrenal axis becomes overactive, the permissive role of cortisol on HSL activation should increase. A foreseeable result of this phenomenon is an enhanced lipolysis. The high efflux of FFAs and glycerol in the blood will increase hepatic gluconeogenesis, thus impairing glucose homeostasis. Under this situation, triglycerides would migrate from the adipocytes (normotopic storage) into lean organs (ectopic storage). Fat relocation should produce a “triglyceride-depleted adipocyte.” On the other hand, fat relocation and ectopic fat accumulation should reduce leptin secretion and induce lipotoxicity in lean organs such as the liver, skeletal muscles, and endocrine pancreas.

5. Conclusions

a) If hypoleptinemia drives adrenal axis hyperactivity in CGL (Figure 3), three interventions (recombinant leptin, anti-glucocorticoids, and bilateral adrenalectomy) should reduce fat exposure to cortisol action in CGL patients (Figure 4). Moreover, on the horizon, the nonpeptide, oral ACTH antagonist CRN04894 [34] might become a promising therapeutic alternative. If ACTH antagonists reach the market and prove safe for chronic use, administration of these drugs in CGL patients may be beneficial due to their adrenal-blocking properties.

Figure 3.

Potential pathophysiological mechanisms underlying the hypoleptinemia-induced hyperactivity of the hypothalamic–pituitary–adrenal axis (adrenal axis, for short) and subsequent unrestrained adipose tissue lipolysis. Since leptin restrains the hypothalamic–pituitary–adrenal axis, any cause of severe hypoleptinemia (CGL, severe hypoinsulinemia of diabetic ketoacidosis, and prolonged fasting) should result in an overactive adrenal axis. Fat exposure to excess serum cortisol should stimulate adipocyte hormone-sensitive lipase (HSL) and increase lipolysis. Abbreviations: FFA, free fatty acids; TG, triglycerides.

Figure 4.

Predicted therapeutic interventions aimed at reducing fat exposure to excessive cortisol levels in patients with Berardinelli-Seip syndrome (congenital generalized lipodystrophy). According to our hypothesis, adrenalectomy (a), anti-glucocorticoid therapy (b), and leptin replacement (c) should result in a restrained activity of the hypothalamic–pituitary–adrenal axis (adrenal axis, for short), reducing fat depots exposure to free cortisol levels. The same effects should result from using future ACTH antagonists (d). Consequently, any of these four interventions should diminish the degree of lipolysis. In turn, reduced lipolysis should ameliorate ectopic fat storage in lean organs (liver, muscle, pancreatic beta cells, and skin), improving tissue insulin sensitivity. Abbreviation: ACTH, adrenocorticotropic hormone; FFA, free fatty acids.

b) Mifepristone is not suitable for patients with CGL, since it induces adrenal axis overactivity. This fact anticipates complications such as adrenal hyperplasia and hypercortisolemia-induced hypermineralocortisolism.

c) Bilateral, total adrenalectomy might become a feasible therapeutic alternative for CGL patients. Currently, laparoscopic adrenalectomy entails a low long-term risk to patients. Adrenalectomized patients are perfectly able to manage their hormone replacement therapy.

d) The exact mechanism by which anti-glucocorticoid therapy resulted in the notable metabolic improvement observed in our patient remains unknown and should be investigated. Simultaneous defects in adipogenesis, lipogenesis, and lipolysis may cause lipodystrophy in patients with CGL. “Runaway lipolysis” by itself as the single culprit of lipodystrophy remains an unproven possibility.

e) The storage of fat in adipose tissue depends on the correct functioning of adipogenesis, lipogenesis, and lipolysis, working as a whole process. It may be plausible that a failure in just one of these three elements is sufficient to derange the whole process, leading to lipodystrophy.

f) An urgent task is to study the status of the adrenal axis in patients with untreated CGL. If this axis turns out to be overactive, future clinical trials with anti-glucocorticoids in patients with CGL are warranted.

Acknowledgments

This author is indebted to several people that facilitated this work. In the first place, thanks are due to Harold Michelsen, MD (deceased), who asked me to develop a therapeutic strategy for this complex patient after the lack of success of the usual medicines to treat her diabetes and dyslipidemia. This author is also grateful to the endocrinology colleagues in our Unit, who were brave enough to endorse our proposed—unheard-of—strategy to attempt a rational treatment of our patient. This author is also grateful to our patient and her family, who trusted our Unit in times of therapeutical uncertainty. They not only accepted the anti-glucocorticoid therapeutic approach but also dared to try bilateral, irreversible adrenalectomy after our metabolic success. Lastly, but not least, this author must thank Inés Vega, RN, and her crew for helping the patient along the road to achieving better health.

Conflicts of interest

The author declares no conflict of interest.