. 2020 Jul 29;156(9):1–10. doi: 10.1001/jamadermatol.2020.1731

Table 2. Performance of Index GEP Tests in Predicting Melanoma Recurrence by Study and Stage of Disease.

Source	No.		Observed RFS rates, % (follow-up, y)		Association between GEP high score and event		Proportion, %
Source	Patients	Observed events^a	GEP low score	GEP high score	HR (95% CI)	P value	Events classified as high risk (follow-up, y)^b	Nonevents classified as low risk (follow-up, y)^b	High-risk patients with event	Low-risk patients without event
Stage I melanoma
DecisionDx
Hsueh et al,²⁵ 2017	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
Greenhaw et al,²⁹ 2018	219	1	NR	NR	NR	NR	0	92	0	>99
Zager et al,²⁷ 2018	264	17	96 (5)	85 (5)	4.01 (1.5-11.5)^c	.007	35; 40 (5)	87; 87 (5)	15	95
Keller et al,²⁶ 2019	96	3	NR	NR	NR	NR	0	95	0	97
Podlipnik et al,²⁸ 2019	44	0	NR	NR	NR	NR	NA	89	0	100
MelaGenix^d
Koelblinger et al,³¹ 2018	88	22	NR	NR	NR	NR	32	77	32	77
Amaral et al,³⁰ 2020	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Stage II melanoma
DecisionDx
Hsueh et al,²⁵ 2017	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
Greenhaw et al,²⁹ 2018	37	12	NR	NR	NR	NR	83	44	42	85
Zager et al,²⁷ 2018	93	39	74 (5)	55 (5)	2.5 (1.1-5.5)^c	.02	77; 76 (5)	43; 40 (5)	49	72
Keller et al,²⁶ 2019	40	14	NR	NR	NR	NR	86	54	50	88
Podlipnik et al,²⁸ 2019	42	7	NR	NR	NR	NR	100	40	25	100
MelaGenix^d
Koelblinger et al,³¹ 2018	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Amaral et al,³⁰ 2020	245	78	76 (5); 73 (10)	58 (5); 46 (10)	NR	NR	76	43	38	79
Stage I-II melanoma
DecisionDx
Hsueh et al,²⁵ 2017	282	14	99 (1.5)	85 (1.5)	NR	NR	79	82	19	99
Greenhaw et al,²⁹ 2018	256	13	98 (3); 93 (5)	74 (3); 69 (5)	OR, 22.0 (5.7-84.2)^e	.01	77; 78 (3); 73 (5)	87; 79 (3); 70 (5)	24	99
Zager et al,²⁷ 2018	357	56	NR	NR	NR	NR	64	79	36	92
Keller et al,²⁶ 2019	136	17	NR	NR	NR		71	86	41	95
Podlipnik et al,²⁸ 2019	86	7	NR	NR	18.8 (1.8-2549.8)^f	.01	100	67	21	100
MelaGenix^d
Koelblinger et al,³¹ 2018	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Amaral et al,³⁰ 2020	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA

Abbreviations: GEP, gene expression profile; HR, hazard ratio; NA, not applicable; NR, not reported; OR, odds ratio; RFS, recurrence-free survival.

^{^a}

Melanoma recurrence and melanoma relapse were considered synonymous. Hsueh et al²⁵ defined recurrence as regional or distant metastasis. Greenhaw et al²⁹ defined recurrence as satellite, in-transit, nodal, or distant metastasis. Zager et al²⁷ defined recurrence as any local, regional, or distant metastasis. Podlipnik et al,²⁸ Keller et al,²⁶ and Koelblinger et al³¹ did not provide a clear definition of disease recurrence. Amaral et al³⁰ defined recurrence as all melanoma-specific disease progressions.

^{^b}

Unless indicated by a particular cross-sectional follow-up time (ie, 3- or 5-year), reported proportions were calculated using the raw number of high-score GEP patients with an event, number of high-score GEP patients without an event, number of low-score GEP patients with an event, and number of low-score GEP patients without an event. If indicated by a cross-sectional follow-up time, these estimates represent the sensitivity and specificity of the test and were estimated from Kaplan-Meier curves and/or data (see Methods).

^{^c}

Not reported in study but estimated by reviewers by extracting data from published Kaplan-Meier curves; therefore, it is not adjusted for any confounding variables.

^{^d}

Koelblinger et al³¹ and Amaral et al³⁰ used different GEP score cut-offs.

^{^e}

Univariate OR was not reported in the study at a particular cross-sectional follow-up time or by stage of disease. The reviewers calculated the univariate OR to be 3.6 (95% CI, 0.1-91.9; P = .44) for patients with stage I disease and 3.9 (95% CI, 0.7-21.7; P = .12) for patients with stage II disease.

^{^f}

Multivariate HR adjusted for American Joint Committee on Cancer stage (IIB-IIC) and age (>50 years).