2.1 Clinical characteristics of the subjects
Clinical and molecular characteristics of these 24 subjects are described in Table 1. Representative photographs of nine isolated macrodactyly patients are shown in Figure 1. 22 of 24 patients displayed asymmetric and disproportionate overgrowth in their hands or feet at birth, while the remaining 2 patients had later onset symptoms between six and twelve months after birth. With the exception of one 34 year old, the majority of patients were under 15 years old, with a median age of 5. A slight male predominance was observed (16 males versus 8 females, binomial P=0.152) in our cohort. 14 patients presented exclusive involvement of the lower extremities, while nine had macrodactyly exclusively in the upper extremities (binomial P=0.405). Only one patient had overgrowth in both the upper and the lower extremities. Twenty of the 24 patients had unilateral involvement, with 12 patients presenting on the right side and eight patients presenting on the left side of the body (binomial P=0.503). The number of affected digits ranged from one to seven, with an average of 2.7 digits affected. The second digit was the most frequently affected digit (N=22), followed by the third (N=17). Over half of our patients (14/24) had two affected digits, and the combination of second- and third-digit overgrowth (N=10) was more frequently observed than the combination of first- and second-digit enlargement (N=4; binomial P=0.180). In addition, no other combinations were observed in patients with two affected digits. 5 of the 24 patients had syndactyly, a condition where two or more digits are fused together, all of which presented as syndactyly of 2-3 toes.
Patients 1 to 4 had variable presentations of cerebriform connective tissue nevi adjacent to overgrown digits. Therefore, diagnoses of Proteus syndrome were established in these patients.
2.2 Genetic characteristics
Through genetic testing and analysis, we achieved molecular diagnoses of all 24 patients. In the 4 patients who met the diagnostic criteria of Proteus syndrome, we identified an AKT1 c.49G>A (p.Glu17Lys) variant, which is the only variant known to cause Proteus syndrome [12]. Variant allele frequencies (VAFs) of this AKT1 mutation in affected tissues ranged from 10% to 22%, with an average variant frequency of 16%. No variant read was identified in blood DNA (Table 1).
In the 20 patients with isolated macrodactyly who did not meet the diagnostic criteria for Proteus syndrome, we identified and confirmed pathogenic variants in PIK3CA (Table 1). The most commonly observed variant was PIK3CA p.His1047Arg (N=7), followed by p.Glu542Lys (N=6), p.Glu545Lys (N=2), p.His1047Leu (N=2), p.Glu453Lys (N=1), p.Gln546Lys (N=1) and p.His1047Tyr (N=1) (Table 1). VAFs in affected tissues ranged from 10% to 33% with an average of 21% (Table 1). None of these variants were identified in peripheral blood samples.
All seven variants have been previously reported to cause developmental disorders [10,11,13,14], and either predicted or validated to have gain-of-function mechanisms. In vivo studies have demonstrated that mutations in PIK3CA are sufficient to induce oncogenic transformation in chicken embryo fibroblasts through enhancement lipid kinase activity and activation of mTOR and AKT1 signaling [15]. However, all seven variants were absent from the Deciphering Disorders Involving Scoliosis and COmorbidities (DISCO, http://discostudy.org/) study composed of 4000 exome sequencing data of the Chinese population [16–18]. PIK3CA p.Glu453Lys, p.Glu542Lys, p.Gln546Lys and p.His1047Tyr were absent from the Genome Aggregation Database (gnomAD, https://gnomad.broadinstitute.org). PIK3CA p.Glu545Lys, p.His1047Leu, p.His1047Arg and AKT1 p.Glu17Lys were present at extremely low frequencies in gnomAD, with an allele frequency of 4e-6. Despite previously reported observations of PIK3CA p.Gln546Lys, p.His1047Tyr and p.Glu453Lys mutations in other PIK3CA-related overgrowth syndromes (PROS) [11,13,19–21], our study represents the first time they are identified in isolated macrodactyly.
The PI3K protein has five functional domains, including PI3K-ABD, PI3K-RBD, C2 PI3K-type, PIK helical and PI3K/PI4K kinase domain. The p.Glu453Lys mutation is located in the C2 domain. Meanwhile, the p.Glu542Lys, p.Glu545Lys and p.Gln546Lys mutations occurr in adjacent amino acids of the helical domain. Lastly, the p.His1047Arg, p.His1047Tyr and p.His1047Leu mutations are located at the kinase domain of PIK3CA (Figure 2). The majority of established functional variants of the PIK3CA cluster were found in the kinase and helical domains [22], which is consistent with our findings. In the 20 patients carrying pathogenic PIK3CA variants, 9 had variants of the helical domain and 10 had patients affecting the kinase domain, and only one patient had a variant affecting the C2 domain.
2.3 Genotype-phenotype correlation
We then analyzed the potential correlation between subjects’ phenotypes (i.e. macrodactyly of the upper or the lower limb, the number of affected digits, with or without syndactyly) and the domain location of variant mutations in patients carrying the PIK3CA variant (Table 2).
All 9 (100%) patients carrying variants in the helical domain of PI3K presented with macrodactyly in the lower limbs. In contrast, only 4 of 13 (36%) patients carrying variants outside of the helical domain (in the C2 domain or kinase domain) had a lower limb affected. These data demonstrates that helical domain variants are enriched in patients with lower limb overgrowth relative to all other variants (P=0.005; determined by Fisher’s exact test). We also compared the VAFs between the two phenotypically different groups, but no significant correlation was observed.
13 patients had one or two affected digits, and five (38%) of these patients carried PIK3CA mutations located in the helical domain. 7 patients had more than two affected digits, 4 (57%) of whom had a PIK3CA mutation in the helical domain. While it may appear at first glance that patients with less than three affected digits were more likely to have a variant not located in the helical domain, and vice versa, this observation did not prove to be statistically significant (P=0.642; determined by Fisher’s exact test).