PLIN1 Haploinsufficiency Is Not Associated With Lipodystrophy.

Context: Monogenic partial lipodystrophy is a genetically heterogeneous disease where only variants with specific genetic mechanisms are causative. Three heterozygous protein extending frameshift variants in PLIN1 have been reported to cause a phenotype of partial lipodystrophy and insulin resistance. Objective: We investigated if null variants in PLIN1 cause lipodystrophy.
Methods: As part of a targeted sequencing panel test, we sequenced PLIN1 in 2208 individuals. We also investigated the frequency of PLIN1 variants in the gnomAD database, and the type 2 diabetes knowledge portal.
Results: We identified 6/2208 (1 in 368) individuals with a PLIN1 null variant. None of these individuals had clinical or biochemical evidence of overt lipodystrophy. Additionally, 14/17,000 (1 in 1214) individuals with PLIN1 null variants in the type 2 diabetes knowledge portal showed no association with biomarkers of lipodystrophy. PLIN1 null variants occur too frequently in gnomAD (126/138,632; 1 in 1100) to be a cause of rare overt monogenic partial lipodystrophy. Conclusions: Our study suggests that heterozygous variants that are predicted to result in PLIN1 haploinsufficiency are not a cause of familial partial lipodystrophy and should not be reported as disease-causing variants by diagnostic genetic testing laboratories. This finding is in keeping with other known monogenic causes of lipodystrophy, such as PPARG and LMNA, where only variants with specific genetic mechanisms cause lipodystrophy.

Lipodystrophies are rare disorders characterized by a paucity of subcutaneous fat typically associated with metabolic complications including insulin resistance, diabetes mellitus (DM), hyperandrogenism in females, fatty liver disease, and dyslipidemia (1–4).

Heterozygous protein extending frameshift variants in PLIN1 have been reported to cause a phenotype of partial lipodystrophy and insulin resistance (1). PLIN1 encodes perilipin1, a lipid droplet coat protein predominantly expressed in adipocytes. Gandotra et al. (1) identified two heterozygous protein-extending frameshift variants in PLIN1 in three families with partial lipodystrophy and insulin-resistant diabetes. Kozusko et al. (5) reported a different heterozygous protein-extending frameshift variant in two families with a similar phenotype.

With the decreasing cost of DNA sequencing, diagnostic genetic testing no longer focuses on screening individual genes but assesses panels of genes or full exomes in a single test. This type of test is extremely useful for genetically heterogeneous conditions such as lipodystrophy. This less selective approach relies on the correct understanding from the literature of which variants within the tested genes are truly pathogenic.

As part of a routine gene panel test for genes reported to be causative of lipodystrophy and monogenic diabetes, we sequenced PLIN1 in 1323 patients referred for Maturity Onset Diabetes of the Young (MODY) testing. We subsequently studied a further 885 patients with neonatal diabetes or hyperinsulinism who were tested using the same targeted gene panel. We also investigated the frequency of PLIN1 null variants in 138,632 people in gnomAD (6) and lipodystrophy and diabetes phenotypes in 17,000 individuals in the diabetes knowledge portal. Only three specific PLIN1 variants have been reported to cause lipodystrophy; our aim was to investigate if null variants in the gene were associated with the same phenotype.

Materials and Methods

Recruitment and phenotyping of cohort

At the Royal Devon and Exeter Hospital, 2208 patients were referred for genetic testing to the Molecular Genetics Laboratory. We initially studied 1323 patients with a clinical suspicion of MODY. Subsequently, we studied a further 410 patients with hyperinsulinemic hypoglycemia and 475 patients with neonatal diabetes. Clinical information was provided on a standardized referral form by the clinician at the time of referral for genetic testing, and additional follow-up was carried out for four families. At follow-up, the patients and family members were assessed based on the multisociety practice guidelines on diagnosing lipodystrophy (7). These guidelines provide the list of clinical features that increase suspicion of lipodystrophy. The list contains the following features: generalized or regional absence of body fat, failure to thrive (infants and children), prominent muscles, prominent veins, severe acanthosis nigricans, eruptive xanthomata, cushingoid/acromegaloid/progeroid appearance, DM with high insulin requirements (≥200 U/d or ≥2 U/kg/d or requiring U-500 insulin), severe hypertriglyceridemia (≥500 mg/dL with or without therapy or ≥250 mg/dL despite diet and medical therapy), history of acute pancreatitis secondary to hypertriglyceridemia, nonalcoholic steatohepatitis in a nonobese individual, early-onset cardiomyopathy, and polycystic ovary syndrome.

An insulin measurement was obtained for the father of proband 1 using an Access Ultrasensitive Insulin assay run on a Beckman Coulter Unicel DXI 800 Access Immunoassay System (normal range, 13.0 to 161). Proband 4 had their insulin measured using an Abbott Architect insulin reagent kit on an Abbott Architect i2000SR (reference <83 pmol/L). Reference ranges for the assays were obtained from the laboratories that performed the assays.

A dual-energy x-ray absorptiometry (DEXA) scan was performed on proband 4. This measures the fat distribution within the body and allows the calculation of a fat mass ratio—the ratio between the fat mass of the trunk and the lower limbs (8).

Genetic analysis

We sequenced PLIN1 (NM_002666.4) by targeted next generation sequencing, as part of a gene panel test (methodology described previously) (9). The panel contains baits for genes in which pathogenic variants are reported to cause a range of pancreatic-linked phenotypes (including lipodystrophy, monogenic diabetes, and hyperinsulinism); only those relevant to the patient’s disease are reported. Each sample was sequenced with a minimum of 3.3 million reads and greater than 99% of targeted bases covered at 30× or higher. All PLIN1 variants were also confirmed by Sanger sequencing.

Results

The phenotype of patients with heterozygous null variants in PLIN1 is not consistent with overt lipodystrophy

We initially tested 1323 patients with a clinical suspicion of MODY of which four had a heterozygous null variant in PLIN1 (probands 1 to 4; Table 1 and Fig. 1). We investigated data from a further 885 non-MODY patients sequenced with the same genetic panel test, two of whom had a heterozygous null variant in PLIN1 (probands 5 and 6; Table 1 and Fig. 1). All of these variants are nonsense or frameshift variants. None of these patients had symptoms of lipodystrophy. The published patients with pathogenic PLIN1 variants had dyslipidemia with hypertriglyceridemia (range from 2 to 147 mmol/L) (1, 5); all of the patients in our study with variants in PLIN1 had triglyceride levels below 1.7 mmol/L. Proband 4, who has a genetic diagnosis of GCK MODY due to a heterozygous pathogenic missense variant (p.Val182Met), had a fasting glucose of 7 mmol/L, fasting insulin of 20.8 pmol/L, and had a DEXA scan showing his fat mass ratio is 0.52. Valerio et al. (8) used DEXA to measure the fat mass ratio of a cohort of patients with partial lipodystrophy and showed that their mean ratio was 1.86 compared with controls with a mean ratio of 0.93. However, although these results demonstrate that the phenotype of proband 4 is inconsistent with lipodystrophy, he is only 10 years of age thus features could present later and DEXA scans at this age cannot conclusively exclude lipodystrophy.

Table 1.

Details of PLIN1 Variants and Phenotypic Data for the Probands and Family Members in Our Cohort

Family ID	1		2	3	4		5			6
Relation	Proband	Father	Proband	Proband	Proband	Father	Proband	Brother	Mother	Proband	Mother
Sex	Female	Male	Female	Female	Male	Male	Male	Male	Female	Female	Female
Age at study, y	12	48	25	31	10	44	1	6	37	1	40
Ethnicity	Asian (Filipino), UK	Asian (Filipino), UK	Mixed (White and Black African), UK	White, UK	White, Australia	White, Australia	Arabic, Saudi Arabia	Arabic, Saudi Arabia	Arabic, Saudi Arabia	White, Austria	White, Austria
c.Nomen	NM_002666.4:c.760del	NM_002666.4:c.760del	NM_002666.4:c.985C>T	NM_002666.4:c.964-1G>A	NM_002666.4:c.433del	NM_002666.4:c.433del	NM_002666.4:c.768_770delinsTG	NM_002666.4:c.768_770delinsTG	NM_002666.4:c.768_770delinsTG	NM_002666.4:c.1448_1466dup	NM_002666.4:c.1448_1466dup
p.Nomen	p.Val254Trpfs*4	p.Val254Trpfs*4	p.Arg329*	p.?	p.Ala145Profs*75	p.Ala145Profs*75	p.Leu257Glufs*39	p.Leu257Glufs*39	p.Leu257Glufs*39	p.Glu490Leufs*82	p.Glu490Leufs*82
Exon	6	6	8	8	5	5	6	6	6	9	9
Diagnosis	DM	Unaffected	Renal agenesis	DM	DM	Unaffected	HI	HI	Unaffected	HI	Unaffected

Abbreviation: HI, hyperinsulinemic hypoglycemia.

Figure 1.

Pedigrees of the families with PLIN1 null variants. ALP, alkaline phosphatase; ALT, alanine transaminase; BMI, body mass index; Dx, diagnosed; GAD, glutamic acid decarboxylase autoantibodies; GCK, glucokinase; HbA1c, glycated hemoglobin; HI, hyperinsulinemic hypoglycemia; HLD, high-density lipoprotein; LD, lipodystrophy; LDL, low-density lipoprotein; NR, normal range; WHO, World Health Organization.

Symptoms of familial partial lipodystrophy typically present at puberty in women. Two of our probands were over 18 years old and did not display the characteristic fat displacement or have biochemical evidence of overt lipodystrophy. Additionally, family member testing for four probands identified the PLIN1 variant in an unaffected parent and one sibling (Table 1). Three of the parents had no overt symptoms of lipodystrophy at the ages of 40, 44, and 46, and although the father of proband 1, 48 years old, showed impaired fasting glycemia with high cholesterol level but normal fasting triglyceride and high high-density lipoprotein levels, these results are not consistent with severe overt lipodystrophy.

Null variants in PLIN1 were not associated with biomarkers of lipodystrophy in an independent cohort

The lack of an overt lipodystrophy phenotype in patients with null variants in PLIN1 raised questions over the pathogenicity of protein truncating variants in this gene. We used the type 2 diabetes knowledge portal to assess the association of PLIN1 to the biomarkers of lipodystrophy (10). The type 2 diabetes knowledge portal is a repository of 17,000 people who have been exome sequenced and had certain phenotypes tested. There are seven different PLIN1 null variants in 14 people, giving a frequency of 1 in 1214. PLIN1 null variants were not associated with higher fasting insulin, higher triglyceride, higher low-density lipoprotein, or lower high-density lipoprotein (Supplemental Table 1), indicating that these individuals do not have overt lipodystrophy.

Null variants in PLIN1 are present at a frequency in gnomAD which is higher than expected

In the publically available gnomAD database, 126/138,632 (1 in 1100) people have heterozygous null variants (6). PLIN1 is not well covered in all individuals in gnomAD, so the true frequency could be even higher (see Supplemental Fig. 1). The pLI constraint score for loss of function intolerance is 0 for PLIN1, predicting that these variants are tolerated based on their frequency (6). These findings are not consistent with heterozygous null variants causing overt lipodystrophy. However, there are no null variants in a homozygous state in gnomAD.

Discussion

PLIN1 variants are unlikely to be pathogenic via a mechanism of haploinsufficiency

We present six probands and five family members who have null variants in PLIN1 but no phenotype of lipodystrophy. Importantly, six of these individuals are over 18 years old, when familial partial lipodystrophy is expected to have presented, and four are over the age of 40 years old. We were able to obtain follow-up information for four families. Although it was not possible to obtain full information on all of these individuals, we have clear biochemical measures, such as triglycerides, for several individuals demonstrating they are unlikely to have severe lipodystrophy at the time of study (however, proband 4 was only 10 years of age; thus, there is the potential to develop symptoms at a later date). Another limitation of our study is that all biochemistry parameters were measured locally with different commercial assays. We have limited information in our original cohorts on all the features of lipodystrophy that have been proposed by the multisociety practice guidelines (7). However, this is not a concern in our study as all of the patients in the cohort had undergone genetic testing for all the genes on our panel, irrespective of their clinical features. This robust analysis has made sure that we did not miss any patients with variants in PLIN1. It is possible that although PLIN1 heterozygous null variants do not appear to cause the overt lipodystrophic phenotype described in Gandotra et al. (1), they may cause a more subtle form of the disease.

Familial partial lipodystrophy is estimated to have a prevalence of approximately 3 in 1 million in the general population (11). The frequency of null variants in the gnomAD database is incompatible with the frequency of lipodystrophy. The prevalence of familial partial lipodystrophy would need to be greater than 1 in 10,000 to be consistent with null variants in PLIN1 causing the disorder (12). If these variants caused familial partial lipodystrophy, then the disease would be expected to affect 60,000 people in the United Kingdom. Additionally, the type two knowledge portal provides a set of individuals with a high frequency of PLIN1 variants where there is no association with raised insulin or lipids. Thus, variants in PLIN1 are unlikely to be acting via a mechanism of true haploinsufficiency, i.e., complete lack of one allele.

PLIN1 is an example of a gene where only variants with specific genetic mechanisms are likely to be pathogenic

There is a range of genetic mechanisms of pathogenicity for the known monogenic causes of lipodystrophy. For example, some variants in PPARG demonstrate dominant negative behavior (2, 3), whereas others appear to be pathogenic due to haploinsufficiency. A different pattern is observed in LMNA where haploinsufficiency causes phenotypes including cardiomyopathy, whereas specific missense variants result in lipodystrophy, and other specific mutations in LMNA cause distinct phenotypes, including progeria (4).

The PLIN1 frameshift variants published as pathogenic do not appear in gnomAD. However, it is also possible that the reason we do not see these in gnomAD is that the last exon has very low sequencing coverage: the majority of samples have less than 10× coverage (see Supplemental Fig. 1). The absence of the published variants in gnomAD and our cohort prevents us from commenting on their pathogenicity.

A possible explanation for the high frequency of PLIN1 null variants in public databases would be that variants are pathogenic but have low penetrance or reduced clinical expressivity. However, given their frequency, they would either need to have very low penetrance or the disease would need to be more common than assumed (more common than 1 in 10,000).

PLIN1 variants resulting in haploinsufficiency should not be reported as pathogenic without further evidence

Doubt over haploinsufficiency in PLIN1 causing lipodystrophy has important consequences for genetic testing. PLIN1 is currently screened on diagnostic targeted gene panel tests for lipodystrophy, insulin resistance, and diabetes. We recommend that null variants in PLIN1 should not be reported as causative of lipodystrophy.

Our study suggests that heterozygous null variants in PLIN1 do not cause overt lipodystrophy and provides another example of a gene where only variants with specific genetic mechanisms cause lipodystrophy.

Click here for additional data file.