Overview

This report compares carriers of the mitochondrial variant m.C2639T (MT-RNR2 / 16S rRNA) against non-carriers across the DOLORisk diabetic peripheral neuropathy cohort.

Carrier identification

Carriers are extracted directly from the mitochondrial VCFs at position 2639, mapped from sequencing (WTCHG) IDs to DOLORisk IDs, and matched to the cleaned phenotype table. 17 carriers are identified (euNeuP 13, eunoNeuP 4) and all match a phenotype record. Following the official QC, the 2 flagged exclude are set aside, leaving 15 in the primary comparison.

17 identified carriers → 15 analysed
Analysis status Carriers
Case — neuropathic pain 11
Control — painless 4
Excluded (DOLORisk criteria) 2
Total — 17

Primary comparison: carrier vs non-carrier

Continuous phenotypes are compared with the Wilcoxon rank-sum test, categorical phenotypes with Fisher’s exact test.

Missing values are preserved, not imputed. Each test uses pairwise deletion, for a given phenotype, records missing that value are dropped for that test only , so the effective sample size varies by phenotype (the n= in each row of the table). No imputation is applied in this primary comparison; imputation is reserved for a separate sensitivity analysis.

The phenotypes tested carry very different amounts of missingness, which sets which are imputation candidates:

Missingness of tested phenotypes (analysis set)
Phenotype Missing % Carrier n Tier
MNSI score 69.2 11 High (≥40%) — describe only
HbA1c (mmol/mol) 53.2 11 High (≥40%) — describe only
BPI average 52.2 14 High (≥40%) — describe only
Diabetes type 41.5 14 High (≥40%) — describe only
TCSS (Toronto) 38.9 14 Impute candidate (5–40%)
DN4 score 35.0 14 Impute candidate (5–40%)
BMI (kg/m²) 21.0 15 Impute candidate (5–40%)
Pain intensity 9.9 15 Impute candidate (5–40%)
Age (years) 0.2 15 Near-complete (<5%)
Sex 0.2 15 Near-complete (<5%)
NPalgo grade 0.2 15 Near-complete (<5%)
Pain status 0.0 15 Near-complete (<5%)
m.C2639T carrier vs non-carrier — all phenotypes
Variable Type Carrier Non-carrier Effect (95% CI)1 p
Age (years) Continuous — median (IQR) n=15 · 66.0 (58.0–69.5) n=2407 · 64.0 (54.0–72.0) Δ 0.0 (-5.0, 6.0) 0.880
BMI (kg/m²) Continuous — median (IQR) n=15 · 29.4 (26.5–33.1) n=1902 · 27.8 (24.5–31.8) Δ 1.6 (-1.1, 4.4) 0.237
HbA1c (mmol/mol) Continuous — median (IQR) n=11 · 54.0 (48.0–66.5) n=1126 · 56.0 (48.0–68.0) Δ -1.0 (-9.0, 7.0) 0.859
DN4 score Continuous — median (IQR) n=14 · 4.0 (3.0–5.8) n=1564 · 5.0 (3.0–6.0) Δ -1.0 (-2.0, 1.0) 0.330
BPI average Continuous — median (IQR) n=14 · 3.0 (1.0–4.5) n=1147 · 4.0 (2.0–6.0) Δ -1.0 (-3.0, 0.0) 0.066
TCSS (Toronto) Continuous — median (IQR) n=14 · 8.5 (7.2–12.0) n=1469 · 10.0 (7.0–13.0) Δ -1.0 (-3.0, 2.0) 0.540
MNSI score Continuous — median (IQR) n=11 · 5.0 (4.0–7.0) n=738 · 5.0 (3.0–7.0) Δ 0.0 (-1.0, 2.0) 0.839
Pain intensity Continuous — median (IQR) n=15 · 2.0 (0.0–4.0) n=2172 · 3.0 (0.0–5.0) Δ -0.0 (-2.0, 1.0) 0.637
Sex Categorical — Fisher n=15 n=2408 OR 0.42 (0.08, 1.57) 0.192
Diabetes type Categorical — Fisher n=14 n=1406 OR 0.00 (0.00, 2.74) 0.385
Pain status Categorical — Fisher n=15 n=2413 OR 1.16 (0.34, 5.03) 1.000
NPalgo grade Categorical — Fisher n=15 n=2408 — 0.183
1 Δ = Hodges–Lehmann median difference (carrier − non-carrier); OR from 2×2 Fisher. '—' where no single effect size is defined (multi-level).

Forest plot with CI also include here as the small sample size we have.

Marginal differences: binned age & sex

First the demographic marginals, whether carrier status itself associates with age band or sex (potential confounders). Then the marginal phenotype differences (carrier vs non-carrier) within each stratum.

Sex-stratified tests

Each phenotype is re-tested within males and within females separately. Note the female carrier stratum has only ~3 individuals, so those p-values are shown for completeness.

Carrier vs non-carrier, stratified by sex
Variable Male n Male p Female n Female p
Age (years) 12 0.643 3 0.487
BMI (kg/m²) 12 0.284 3 0.738
HbA1c (mmol/mol) 8 0.699 3 0.372
DN4 score 11 0.582 3 0.363
BPI average 11 0.503 3 0.023
TCSS (Toronto) 11 0.294 3 0.734
MNSI score 9 0.744 2 0.856
Pain intensity 12 0.825 3 0.222
Diabetes_type 11 0.612 3 1.000
Pain_status 12 0.761 3 1.000
N_palgo 12 0.394 3 0.491

NPalgo extremes: grade 0 vs grade 4

Restricting the neuropathic-pain axis to : grade 0 (no neuropathic pain) vs grade 4 (definite neuropathic pain) .

Robustness: exact tests & FDR correction

I also quick check with the 20,000 permutation and across the 12 primary tests we control the false discovery rate (Benjamini–Hochberg q-value).

Exact tests and FDR-adjusted significance
Phenotype Method p (primary) p (exact/perm) q (BH-FDR)
BPI average Permutation (Wilcoxon) 0.066 0.065 0.711
NPalgo grade Fisher (exact) 0.183 0.183 0.711
Sex Fisher (exact) 0.192 0.192 0.711
BMI (kg/m²) Permutation (Wilcoxon) 0.237 0.237 0.711
DN4 score Permutation (Wilcoxon) 0.330 0.330 0.769
Diabetes type Fisher (exact) 0.385 0.385 0.769
TCSS (Toronto) Permutation (Wilcoxon) 0.540 0.544 0.933
Pain intensity Permutation (Wilcoxon) 0.637 0.645 0.966
MNSI score Permutation (Wilcoxon) 0.839 0.842 0.966
HbA1c (mmol/mol) Permutation (Wilcoxon) 0.859 0.861 0.966
Age (years) Permutation (Wilcoxon) 0.880 0.885 0.966
Pain status Fisher (exact) 1.000 1.000 1.000