Dilated and arrhythmogenic cardiomyopathy

Gene: PRDM16

Green List (high evidence)

Review of PRDM16 Gene Variants as a Cause of Dilated Cardiomyopathy and Left Ventricular Non-Compaction

Short Summary of Findings

The potential role of this variant in cardiomyopathy-causality was first noticed due to an apparently higher incidence of cardiomyopathy (mainly LVNC, but also DCM) in patients with 1p36 deletion syndrome in whom the deletion includes the PRDM16 gene, as compared to patients with the same syndrome in whom the gene is preserved. This finding has now been replicated in multiple cohorts (e.g. PMIDS: 23768516, 24454898, and 37395136). There have been a quite a significant number of reports (reviewed below) of PRDM16 truncating variants (including multi-exon deletions, nonsense, frameshift and splice site variants) being detected in probands with non-syndromic cardiomyopathy of varying severity and age at onset (mainly LVNC and DCM, a few with features of HCM), and many of these appear to have arisen de novo (a few with parentage confirmed). In contrast, loss-of-function truncating variants are very rare on gnomAD, which indicates significant population constraint for this type of variant, and studies have shown that PRDM16 truncating variants are significantly enriched in cardiomyopathy cohorts as compared to population-based cohorts (e.g. PMIDs: 33500567 and 40935858). Of note, however, there has been very little cosegregation data reported for any PRDM16 variant with cardiomyopathy (two segregations in PMID: 40935858 are the only ones I could find).

There have been multiple knockout and 1 knock-in mouse models which all support the notion that loss-of-function of PRDM16 (mainly biallelic LoF) can cause features of non-compaction and dilated cardiomyopathy. The different models show very different severity in the mouse phenotypes. Of note, only 1 model (a knock-in of a nonsense variant; PMID: 38113297) was noted to cause a cardiac phenotype in the heterozygous state.

Overall, despite the lack of cosegregation data for any reported PRDM16 variant, I think there is fairly strong evidence that truncating variants in this gene are causative of LVNC and DCM. I am not aware of any PRDM16 variant that could currently be classified as ‘pathogenic’, and this is in large part due to the lack of cosegregation data. Also, because no one PRDM16 variant can be classified as ‘pathogenic’ without application of PVS1, then, as per Tayoun et al. (2018; PMID: 30192042) guidelines, I think that PVS1 cannot currently be applied for any such variant any higher than at ‘strong’ level. This means that PRDM16 truncating variants can easily be classified as likely pathogenic by default (assuming PM2 applies, as well as PVS1_strong), which could then potentially reclassified to pathogenic with additional evidence (which may include demonstration of de novo status, and cosegregation data etc.). In conclusion, I agree that the PRDM16 gene can be rated as a ‘green’ gene for LVNC/DCM and paediatric cardiomyopathy.

I have included notes below on all the databases / journal articles I have reviewed. Please note that I have only reviewed the gene with respect to loss-of-function / truncating variants (not missense variants), and my review did not include inspection of ClinVar entries.


Background Information

PRDM16 encodes a transcription factor with a DNA-binding zinc finger domain. It is involved in regulation of hematopoiesis, palatogenesis, neurogenesis, and brown fat development. It has a role in TGFβ signalling. It has been found to have a role in cardiomyocyte proliferation. PRDM16-deficient mice have ventricular hypoplasia.

This gene maps to 1p36, which is the chromosomal region deleted in 1p36 deletion syndrome, which presents with craniofacial dysmorphism, structural brain abnormalities, seizures, hearing loss, and developmental delay. ~25% of individuals with 1p36 deletion syndrome also exhibit cardiomyopathy (usually LVNC, sometimes DCM).


Population Databases

Database of Genomic Variants: There are still a fair number of exon-scale deletions listed in the database that have been detected at fairly high frequency in some studies. For example, a deletion of half of the gene is detected in 16 of 95 samples in Wong et al. (2007; PMID: 17160897), a deletion of multiple 3’ exons is detected in 8 of 443 samples in Jakobsson et al. (2008); PMID: 18288195). The apparently high frequency of exon-scale deletions in DGV was the main reason it was decided not to include PRDM16 in cardiomyopathy gene panels at Oxford lab in 2015.

gnomAD (SVs v4.1.0 and CNVs v4.1.0): A few exonic deletions are listed; however, these are mainly very rare (detected in a single allele). One exception is NM_022114.4(PRDM16):c.3697-107_3722del, which deletes a 5’ portion of the terminal exon, which looks to be very frequent; however, the genotype quality is very low in the majority of the variant alleles, which I think suggests that it could be an artefact, even though it passed quality filters. Overall, LoF variants are indicated to be subject to significant constraint (21 SNVs observed vs 99.4 expected).

So, given the absence of frequent and plausible exonic deletions in gnomAD, I am tending to think that the early array findings of frequent deletions listed in DGV are probably artefactual.


Studies on 1p36del Syndrome Cases

Arndt et al. (2013; PMID: 23768516):
• Identified 18 1p36del syndrome patients with cardiomyopathy, and determined a common minimal ~130 kb region of loss that mapped exclusively to the PRDM16 gene.
o 17 of the 18 individuals had PRDM16 deletion.
o PRDM16 was not deleted in one individual (? phenocopy).
o A later response to the paper points (de Leeuw & Houge, 2014; PMID: 24387995) out that at least one other case also did not have a deletion of PRDM16 (error in given coordinates).
• Then sequenced PRDM16 in 75 Western European non-syndromic LVNC probands and detected three variants:
o De novo p.Lys702*: Detected in affected proband. Not detected in either parent (both unaffected).
o De novo p.Arg525fs: Detected in affected proband. Not detected in either parent or sibling (all unaffected).
• No segregation analysis described.
• Zebrafish with either down-regulation of PRDM16 or transgenic for the p.Lys702* mutation showed cardiac contractile dysfunction and reduced cardiomyocyte proliferation.

Zaveri et al. (2014; PMID: 24454898):
• Similar approach to Arndt et al. (2013) using 15 1p36 del cases with cardiomyopathy.
• Also highlight the PRDM16 region (deleted in 11 cases), but also highlight a second (non-overlapping) cardiomyopathy critical region (deleted in 4 cases).

Kramer et al. (2023; PMID: 37395136): This group found that 10 of 28 PRDM16-deleted cases had a diagnosis of cardiomyopathy (any type) vs just 1 of 11 PRDM16-preserved cases. This difference was noteworthy, but not statistically significant due to lack of statistical power. To increase statistical power, they combined their cohort with cases described in the literature. In this combined cohort, 23 of 79 PRDM16-deleted cases had cardiomyopathy (mainly non-compaction cardiomyopathy but some with DCM) vs 4 of 37 PRDM16-presevered cases. This was statistically significant (p=0.03).


Small-Scale Sequence Variants in Cardiomyopathy Probands

Mazzarrotto et al. (2021; PMID: 33500567): They found a 1.4% excess burden of PRDM16 truncating variants in LVNC cases vs gnomAD controls (which was statistically significant). Six such variants in total (most previously described in other cohorts) detected in 6 of 444 LVNC probands. Odds Ratio for case excess was 235.1 (95% CI: 78.7-702.4). No family testing mentioned.

Long et al. (2017; PMID: 29367541): Whole exome sequencing detected a PRDM16 frameshift variant in a 4-month-old with DCM. The variant was not detected in parents. I assume parentage was confirmed, as parents were also tested by exome sequencing. No segregation in other affected family members is mentioned.

Piekutowska-Abramczuk et al. (2022; PMID: 35893073): Describe a nonsense and a frameshift variant in two paediatric cases of LVNC. In both cases testing of parents indicated they had arisen de novo. Whether or not parentage was confirmed is not indicated. No segregation in other affected family members is mentioned.

van Waning et al. (2018; PMID: 29447731): PRDM16 LoF variants were detected in three adults with non-compaction cardiomyopathy. In one case, the splice variant was said to have arisen de novo. Whether or not parentage was confirmed is not indicated. No segregation in other affected family members is mentioned.

van Lint et al. (2019; PMID: 30847666): Two PRDM16 splice site variants detected in two cardiomyopathy probands. No family testing mentioned.

Ross et al. (2020; PMID: 33082984): A PRDM16 frameshift variant detected in an adult proband with LVNC. No family testing mentioned.

Hirono et al. (2020; PMID: 32183154): A PRDM16 frameshift variant detected in an proband with LVNC. No family testing mentioned.

Schultze-Berndt et al. (2021; PMID: 34540771): Three PRDM16 truncating variants detected in three adult LVNC probands. Two of them are said to be de novo. Whether or not parentage was confirmed is not indicated. No segregation in other affected family members is mentioned.

Wang et al. (2022; PMID: 34350506): A nonsense PRDM16 variant detected in an 8-month-old female with DCM. Said to be de novo. Whether or not parentage was confirmed is not indicated. No segregation in other affected family members is mentioned.

Khan et al. (2022; PMID: 34935411): A frameshift variant (same variant ad detected in one of the Schultze-Berndt cases) was detected in a paediatric DCM proband with no family history. No family testing mentioned.

Sun et al. (2023; PMID: 38113297): Found a PRDM16 nonsense variant and a PRDM16 splice site variant in two cases of paediatric LVNC / DCM. The nonsense variant was not detected in parents indicating it had arisen de novo. Whether parentage was confirmed or not was not mentioned. In both families, an unaffected sibling of the proband was tested and found not to have the variant.

Delplancq et al. (2020; PMID: 31965688): Describe a fetal case of LVNC with hydramnios and an expanded hypokinetic heart. Singleton exome sequencing detected a PRDM16 nonsense variant, which was not detected in the mother, so inferred to be de novo. A de novo TTN missense variant and a maternally inherited +5 TTN intronic variant (inconsistently predicted to affect splicing by different programs) were also detected.

Billon et al. (2025; PMID: 40935858): Three French labs undertook NGS panel testing (59 genes) on 4,900 probands with DCM and / or hyper-trabeculation. In 9 families, heterozygous LoF PRDM16 variants were detected in 11 cardiomyopathy patients. This included 2 whole gene deletions (1 of which described as de novo), 1 multi-exon deletion, and 6 (1 de novo) small-scale truncating variants. 10 of the 11 patients presented with DCM. Two case had features of HCM (in combination with either DCM or hypertrabeculation). Of note is that this paper demonstrates segregation with cardiomyopathy in two affected family members (this is the only paper I have come across demonstrating any cosegregation with cardiomyopathy in family members).


Mouse Models

Homozygous global PRDM16 knockout mice are non-viable (Aguilo et al. 2011; PMID: 21343612). However, mice where the homozygous knockout is restricted to the heart do survive.

Nam et al. (2016; PMID: 32083975) created a cardiac-specific PRDM16 knockout mouse model and examined cardiac function. They did not describe any significant difference between heterozygous cardiac knockout mice vs wildtype mice, but homozygous cardiac knockout mice were observed to have increased QRS duration and QTc interval, and their hearts showed greater fibrosis and cardiomyocyte hypertrophy, which are phenotypes associated with cardiomyopathy. However, there is no mention of these mice developing ventricular dilatation or non-compaction.

Cibi et al. (2020; PMID: 33086060) have also generated cardiac-specific PRDM16 knockout mice, and have found that the gene is dispensable for cardiac development and that adult homozygous knockout mice show no signs of cardiac defects or hypertrophy. However, at ‘advanced age’ (>12 months), the homozygous knockout mice had larger hearts consistent with cardiac hypertrophy, and developed increased fibrosis, left ventricular dilation, and progressive cardiac dysfunction as evidenced by decreased ejection fraction and other parameters, when compared to control mice.

Kramer et al. (2023; PMID: 37395136) also generated cardiac-specific PRDM16 knockout mice and assessed them for sex-dependent effects on cardiac function. They found that at 4 months of age, female knockout mice (I assume these are homozygous), had 38% reduction in ejection fraction and showed signs of left ventricular dilation as compared to controls. In contrast, no such differences were observed in male mice of the same age. However, by 6-7 months of age, both male and female knockout mice displayed a significant decrease in ejection fraction and systolic function. Histologic analysis in 7 month old male and female mice showed significantly more fibrosis in the knockout mice vs controls.

Wu et al. (2022; PMID: 34915728) also generated cardiac-specific PRDM16 knockout mice. In this model, the homozygous mice all died by 7 days of age. The mice were noted to have biventricular non-compaction and dramatic left ventricular dilation.

Sun et al. (2023; PMID: 38113297) found p.Gln187* in a proband with infant-onset LVNC and heart failure. Homozygous knock-in mice with this variant demonstrated an underdeveloped compact myocardium and were embryonically lethal. Heterozygous mice demonstrated significantly smaller ventricular dimensions, heightened fibrosis, and age-dependent loss of TGF-β expression.

I am no expert in evaluating mice models of cardiomyopathy. There seems to be significant differences in the severity of phenotypes between the different mouse models, but I think these studies overall show that PRDM16 loss of function is associated with a cardiac phenotype that recapitulates features of DCM and non-compaction. Of note, with the exception of the mouse model by Sun et al. (2023), most of these models do not demonstrate any harmful effect of PRDM16 haploinsufficiency; however, I am not sure how closely heterozygous mice were examined in these studies, as the reports mainly focus on homozygous mice.

Created: 25 Sep 2025, 2:57 p.m. | Last Modified: 25 Sep 2025, 2:57 p.m.

Panel Version: 3.2

Mode of inheritance
MONOALLELIC, autosomal or pseudoautosomal, NOT imprinted

Phenotypes
LVNC; DCM

Publications

• PMID: 23768516
• PMID: 24454898
• PMID: 37395136
• PMID: 33500567
• PMID: 29367541
• PMID: 35893073
• PMID: 29447731
• PMID: 30847666
• PMID: 33082984
• PMID: 32183154
• PMID: 34540771
• PMID: 34350506
• PMID: 34935411
• PMID: 38113297
• PMID: 31965688
• PMID: 40935858
• PMID: 21343612
• PMID: 32083975
• PMID: 33086060
• PMID: 34915728

Variants in this GENE are reported as part of current diagnostic practice

Green List (high evidence)

The rating of this gene has been updated to green following NHS Genomic Medicine Service approval.

Created: 6 Dec 2024, 9:37 a.m. | Last Modified: 6 Dec 2024, 9:37 a.m.

Panel Version: 2.35

Mode of inheritance
MONOALLELIC, autosomal or pseudoautosomal, imprinted status unknown

Green List (high evidence)

At least six PRDM16 variants have been associated with OMIM:615373 in unrelated cases. See the literature review from Zornitza Stark (https://panelapp.genomicsengland.co.uk/panels/47/gene/PRDM16/)

Created: 4 Jun 2024, 9:54 a.m. | Last Modified: 4 Jun 2024, 9:54 a.m.

Panel Version: 2.29

Green List (high evidence)

Please see review by Zornitza Stark at https://panelapp.genomicsengland.co.uk/panels/47/gene/PRDM16/

Gene is now OMIM Morbid for DCM and Left ventricular noncompaction: https://www.omim.org/entry/615373https://www.omim.org/entry/615373

Created: 15 May 2024, 2:17 p.m. | Last Modified: 15 May 2024, 2:17 p.m.

Panel Version: 2.25

Mode of inheritance
MONOALLELIC, autosomal or pseudoautosomal, NOT imprinted

I don't know

New gene submitted on behalf of the GMS Cardiology specialist group. The group has agreed that this gene should be Amber on this panel.

Created: 3 Dec 2019, 2:28 p.m. | Last Modified: 3 Dec 2019, 2:28 p.m.

Panel Version: 0.52