low heritability for height in subjects with diabetes

Dear GCTA users,

I tried to estimate the heritability for some traits in subjects with diabetes. The sample size is ~5500 and  ~1.2 million SNPs, I obtained the H2 close to 0.2 (SE=0.07) with adjustment for PCA 20, which  is obviously lower than the reported values. For BMI, it seems reasonable (H2=0.4; SE=0.07).  Is it somehow due to the type of population, because all subjects are patients with diabetes? Any comments about it? Many thanks.


#Heritability for height (non-adjustment)
Source Variance SE

V(G) 0.002075 0.000449

V(e) 0.004946 0.000448

Vp 0.007021 0.000134

V(G)/Vp 0.295603 0.063306

logL 10863.667

logL0 10843.656

LRT 40.022

df 1

Pval 1.256e-10

n 5487


#Heritability for height (adjustment for PCA20)
Source Variance SE

V(G) 0.001413 0.000498

V(e) 0.005569 0.000499

Vp 0.006982 0.000134

V(G)/Vp 0.202417 0.070882

logL 10837.413

logL0 10833.051

LRT 8.723

df 1

Pval 0.001571

n 5487


#Heritability for BMI (adjustment for PCA20)
Source Variance SE

V(G) 6.319351 1.173460

V(e) 9.401788 1.142994

Vp 15.721139 0.303253

V(G)/Vp 0.401965 0.073261

logL -10239.846

logL0 -10255.100

LRT 30.509

df 1

Pval 1.661e-08

n 5484