. 2018 Jun 18;2(7):806–816. doi: 10.1210/js.2018-00120

Table 2.

Penetrance Estimates for Recurrent PPGL-Associated SDHA Variants

	SDHA Variant Penetrance (%)				Cumulative LOF SNV Penetrance (%) ^a
	Arg31Ter	Arg75Ter	Arg512Ter	Arg585Trp ^b	Cumulative LOF SNV Penetrance (%) ^a
Combined PPGL cohort^c (n = 1959)
vs GnomAD global	0.90 (0.47–1.69)	0.15 (0.01–0.86)	0.15 (0.00–1.59)	0.58 (0.03–5.47)	0.64 (0.39–1.01)
vs GnomAD European^d	0.46 (0.24–0.88)	0.09 (0.01–0.58)	0.08 (0.00–0.92)	0.40 (0.02–5.07)	0.39 (0.23–0.63)
vs GnomAD Genomes^e	0.59 (0.21–1.73)	0.20 (0.01–5.56)	0.05 (0.00–0.10)	0.39 (0.00–36.0)	0.47 (0.22–1.05)
vs ExAC non-TCGA^f	1.34 (0.55–3.34)	0.07 (0.01–0.43)	0.11 (0.00–1.68)	0.45 (0.02–7.33)	0.60 (0.33–1.06)
Bausch et al. (16) PPGL cohort (n = 972)
vs GnomAD global	0.31 (0.08–0.93)	0.30 (0.02–1.72)	—	—	0.34 (0.15–0.70)
vs GnomAD European	0.16 (0.03–0.48)	0.19 (0.01–1.17)	—	—	0.20 (0.09–0.44)
vs GnomAD Genomes	0.20 (0.03–0.96)	0.40 (0.01–10.6)	—	—	0.25 (0.08–0.72)
vs ExAC non-TCGA	0.45 (0.09–1.86)	0.14 (0.01–0.85)	—	—	0.31 (0.12–0.73)
van der Tuin et al. (17) PPGL cohort (n = 393)
vs GnomAD global	3.38 (1.69–6.41)	—	0.72 (0.01–7.43)	2.85 (0.16–22.3)	1.77 (0.99–3.05)
vs GnomAD European	1.74 (0.86–3.43)	—	0.40 (0.01–4.44)	1.97 (0.09–21.0)	1.08 (0.58–1.92)
vs GnomAD Genomes	2.21 (0.78–6.58)	—	0.25 (0.00–4.78)	1.93 (0.04–73.9)	1.33 (0.56–3.15)
vs ExAC non-TCGA	4.9 (1.96–12.1)	—	0.56 (0.01–7.86)	2.20 (0.09–28.2)	1.66 (0.83–3.18)
vs van der Tuin et al. control cohort^g	1.91 (0.57–7.34)	—	—	—	—

Penetrance estimates are expressed as percent (95% CI). The methods for calculating penetrance estimates together with the respective CIs are described in the Materials and Methods. Baseline lifetime risk of PPGL was estimated to be 0.025% (i.e., 1/4000) based on a midrange incidence estimate of 3 to 3.5/1,000,000 and a ~80-year window of disease susceptibility. Additional details are provided in the footnotes to Supplemental Table 3. The absence of a penetrance estimate (marked —) indicates the absence of the SDHA variant in the respective case and/or control cohort or insufficient information to establish control allele frequencies.

^{^a}

The LOF single-nucleotide variant (SNV) penetrance estimate accounts for the cumulative frequencies of all nonsense and canonical splice site SDHA variants in the respective disease and control cohorts.

^{^b}

The CIs associated with the penetrance estimates for the Arg585Trp variant are noted to be very wide. Notably, this variant was associated with very low variant allele counts (i.e., one or two) in disease and/or control subpopulations, giving rise to large 95% binomial exact CIs for the respective case and control allele frequencies.

^{^c}

Combined cohort as described in Supplemental Tables 2 and 5.

^{^d}

GnomAD European cohort selected to represent most suitable comparator group as each of the combined and individual PPGL cohorts included individuals of predominantly European origin.

^{^e}

GnomAD Genomes cohort was used to reduce the potential for any confounding from the inclusion of samples for The Cancer Genome Atlas (TCGA). Although GnomAD contains 7208 samples from the TCGA database, none are represented by the 15,496 individuals in whom whole-genome sequencing was undertaken (personal correspondence from GnomAD curators).

^{^f}

Exome Aggregation Consortium (ExAC) non-TCGA cohort provides an alternative comparator group in which all TCGA samples (n = 7601) have been removed from the ExAC population, leaving a remaining cohort of 53,105. This was used to establish population allele frequencies with reduced susceptibility to confounding from the inclusion germline samples from individuals with cancer.

^{^g}

The series reported by van der Tuin et al. (17) reported an “in-house” whole-exome control population in which the frequency of the Arg31Ter variant was established. However, no data were provided on other LOF alleles to allow additional allele frequencies to be established.