FCD.evidencereview.092818.4x4.SUBMITTED


Data from 3 studies and 2 physician surveys indicate that the 31-GEP test results significantly
impact management decisions for approximately 1 of 2 patients10-14

Technical success studies demonstrate 99% 
inter- and 100% intra-assay concordance9, 15

Evidence supports consistent ability of the 31-GEP test to accurately identify recurrence,
metastasis, and melanoma-specific mortality in CM patients1-8

Sample adequacy

Literature review of a prognostic 31-gene expression profile (31-GEP) test 
for cutaneous melanoma (CM) risk prediction 

Robert W. Cook, PhD, Kristen M. Plasseraud, PhD, Sarah J. Kurley, PhD, Kyle R. Covington, PhD, Federico A. Monzon, MD, FCAP
Castle Biosciences, Inc., Friendswood, TX

SYNOPSIS & OBJECTIVE

In several cancers, molecular testing has added
prognostic value and utility in the clinical setting. A
31-gene expression profile (31-GEP) test has been
developed and validated for determining metastatic
risk in cutaneous melanoma, with Class 1 and 2
results indicating low and high risks, respectively. As
melanoma staging and guideline recommendations
continue to evolve, it is important to consider the
evidence supporting the use of clinicopathologic and
molecular factors in melanoma patient care. Herein,
published evidence supporting the 31-GEP test,
including clinical validity, analytical validity, and
clinical utility, are reviewed. From clinical validity
evidence spanning eight peer-reviewed articles
(n=1268 total patients) including two prospective
studies, the 31-GEP test consistently demonstrated
accuracy to identify patients with CM at high risk for
recurrence, metastasis, and melanoma-specific
mortality. Published analytical validity data verified
the reliability of 31-GEP testing with inter- and intra-
assay concordance of 99% and 100%, respectively,
and 98% technical success on specimens with
sufficient tumor content. Clinical utility data from
three studies (n=494 total patients) and two
physician surveys indicate that the 31-GEP test
results significantly impact management decisions
for approximately 1 of 2 patients, consistent with the
impact of genomic testing in other cancers. In
contrast to other prognostic melanoma GEP tests
that have been reported, the 31-GEP test has
published evidence from multiple retrospective and
prospective clinical validity studies beyond initial
development, along with published analytical validity
and clinical utility data, in support of its use for
melanoma risk assessment and patient management
decisions.

BACKGROUND & METHODS

•The 31-GEP test is performed in a CAP-
accredited/CLIA-certified laboratory using
high-throughput RT-PCR assays as
previously described1-4.

•Clinical validity, analytical validity, and clinical
utility studies surrounding the 31-GEP are
reviewed herein.

This study was sponsored by Castle Biosciences, Inc., which provided
funding to contributing centers for tissue and clinical data retrieval. RWC,
KPM, SJK, KRC, & FAM are employees and options holders of Castle
Biosciences, Inc.

CLINICAL VALIDITY

CLINICAL UTILITY

ANALYTICAL VALIDITY

REFERENCES

FUNDING & DISCLOSURES

CONCLUSION

Case Clinical Features

1 0.6mm, 42y female, Stage IA

2 0.54 mm, 62y male, Stage IB, ulcerated, personal hx

3 0.76 mm, 69y male, Stage IB, ulcerated, personal hx

4 >0.5mm, 45y male, Stage IB/IIA

5 0.9 mm, 61y female, Stage IB, ulcerated, mitoses

6 1.2 mm, 38y female,  Stage IIA, ulcerated

Does 31-GEP testing alter management decisions and if so, for what modality?13

0

20

40

60

80

100

0.26 0.5 0.76 2.1

%
 R

ec
om

m
en

di
ng

 3
1-

G
E

P

Breslow Thickness (mm)

Percentage of dermatologists who would order the 
31-GEP test in different clinical scenarios

Baseline Ulcerated SLN Negative

*

**
* **

**p<0.05 *p<0.001

Adapted from Svoboda et al.

What features prompt physicians to recommend testing?14

Adapted from Farberg et al.

*p<0.05 Fisher’s 
exact test*

*

*

Figure 6. Schematic representation of using AJCC staging with 31-GEP test result to guide clinical management

High Risk
Class 2

Consider increased 
surveillance intensity

Consider decreased 
intensity

Continue low intensity
managementLow Risk

Stage I-IIA

Early detection 
of recurrence

Melanoma Staging 
(per NCCN)

Low Risk
Class 1

High Risk
Stage IIB-III High Risk

Class 2

Low Risk
Class 1

Continue high
intensity management

Appropriate 
treatment or 
clinical trial

AJCC Staging 31-GEP Test Management Decision

Recommendation changes with GEP result in patient vignettes

Technical reliability 

Sufficient tumor content

Insufficient content (<40%)

96.3%
n=16,472

98.3%
n=16,190

3.7%
n=630

1.7%
n=282

Successful GEP

Multi-gene failure 

Figure 8. The 31-GEP test has high technical reliability on
>17,000 clinical cases since July 201615

Figure 7. 31-GEP results are highly concordant between
assays (n=168 cases)9

Experiment 1 Probability Score

E
xp

er
im

en
t 2

 P
ro

ba
bi

lit
y 

S
co

re

y = 0.9568x + 0.0115
R2 = 0.962

Figure 2. The 31-GEP test is an independent predictor of risk in a
multivariate analysis across a large retrospective cohort of Stage I-III cases 4

1. Gerami P, et al. Clin Cancer Res 2015;21:175-83.
2. Gerami P, et al. J Am Acad Dermatol 2015;72:780-5 e783.
3. Zager JS, et al. BMC Cancer 2018;18(1):130.
4. Gastman BR, et al. JAAD 2018; doi: 10.1016/j.jaad.2018.07.028.
5. Hsueh EC, et al. J Hematol Oncol 2017;10:152.
6. Greenhaw B, et al. Dermatol Surg 2018;

DOI: 10.1097/DSS.0000000000001588.
7. Renzetti M, et al. Society of Surgical Oncology Annual Meeting 2017.
8. Hsueh EC, et al. J Clin Oncol 2016; 34(15_suppl):9565.
9. Cook RW et al. Diagn Pathol 2018;13(1):13.
10. Berger AC et al. Curr Res Med Opin 2016;32(9):1599-604.
11. Dillon LD et al. SKIN: J Cut Med 2018;2(2):111-21.
12. Schuitevoerder D et al. J Drugs Dermatol 2018;17(2):196-199.
13. Farberg AS et al. J Drugs Dermatol 2017;16(5):428-431.
14. Svoboda RM et al. J Drugs Dermatol 2018;17(5):544.
15. Berg AS et al. DERM2018 Conference 2018

Design (n) GEP Impact
Prospectively tested patients, Retrospective 
chart review; (156 patients)10 53%
Prospective documentation 
of pre and post test plans; (247 patients)11 49%
Prospectively tested patients, Retrospective 
chart review; (90 patients)12 52%
Physician survey of clinical decisions with or 
without test results; (169 physicians)13

47-
50%

Physician survey of clinical factors that 
affect use of 31-GEP test; (181 physicians)14 *
*overall GEP impact not assessed with study design

Table 1. Comparison of Clinical Utility Studies 

Figure 5. Physician survey studies address key questions for 31-GEP use13,14

Cook et al.

In review of the literature, the value
of the 31-GEP test for use in
prognosis and clinical management
decision making is supported by
evidence from the 3 pillars of
molecular tests: clinical validity,
clinical utility, and analytical validity.

Cox
Multivariate 

Analysis

RFS DMFS MSS

HR 95% CI P-value HR 95% CI P-value HR 95% CI P-value

Breslow Depth 1.21
1.12-
1.3

<0.0001 1.19
1.09-
1.29

<0.0001 1.16
1.0-
1.34

0.05

Mitotic rate 1.01
0.99-
1.03

0.18 1.01
0.99-
1.03

0.24 0.97
0.92-
1.03

0.34

Ulceration 1.1
0.75-
1.59

0.64 1.57
1.02-
2.43

0.04 0.77
0.38-
1.57

0.47

Positive node 2.45
1.74-
3.46

<0.0001 3.02
2.0-
4.57

<0.0001 3.83
1.85-
7.95

0.0003

Class 1B 1.13
0.56-
2.29

0.73 1.35
0.58-
3.15

.48 4.37
0.84-
22.72

0.08

Class 2A 1.48
0.77-
2.84

0.24 1.53
0.68-
3.43

.30 2.52
0.42-
15.2

0.31

Class 2B 2.92
1.7-
5.00

<0.0001 2.89
1.49-
5.62

0.002 9.02
2.02-
40.24

0.004

HR, hazard ratio; RFS, recurrence-free survival; DMFS, distant metastasis-free survival; MSS, melanoma-specific
survival; GEP, gene expression profile; CI, confidence interval; 147 recurrences, 107 distant metastases, 36
melanoma-specific deaths

Subgroup analysis of this cohort also demonstrated independent ability of 
the 31-GEP test to detect patients at high risk for metastasis in low-risk 

populations of sentinel lymph node-negative, Stage I-IIA, and T1 tumors.  

Figure 1. Accuracy metrics of the 31-GEP test
within a large retrospective cohort (n=690)

•The 31-GEP test predicts a CM patient’s risk
of recurrence, metastasis, or melanoma-
specific mortality at 5 years after diagnosis

Class 1
Low risk of melanoma 

recurrence within 5 years

Class 2:
High risk of melanoma 

recurrence within 5 years

1A 
Lowest risk

2B 
Highest risk

2A
Increased risk

1B  
Low risk

RNA isolation

cDNA generation and amplification (14X)

Open Array PCR gene card 
28 discriminant gene targets and 3 control genes

Patients with 
Stage I-III melanoma
Primary CM tumor tissue

Formalin Fixed, Paraffin Embedded 
(≥ 40% tumor content)

Analysis of gene expression profile with 
a proprietary algorithm to determine 

Class and metastatic risk

Change StudyBerger et al.   Dillon et al.

Class 1 Changed 37% 36%

Changed Class 1 
w/ decrease 94% 67%

Class 2 Changed 77% 85%

Changed Class 2 
w/ increase 94% 92%

Figure 4. 31-GEP result drives surveillance changes in multicenter studies10,11

Consistent specific modality 
changes across both studies:

• Decreases in imaging, 
visits, and referrals with 
Class 1 result

• Increases in labs, imaging, 
visits and referrals with 
Class 2 result

Multi-center study5

n=322%
 R

ec
ur

re
nc

e 
Fr

ee

Figure 3. 31-GEP Class divides patients into high and low-risk categories for recurrence, metastasis and death across multiple
prospective studies5,6,8

Class 1

Class 2

p<0.0001

Single-center study6

p<0.0001

Time (years)

P
ro

ba
bi

lit
y 

A
vo

id
in

g 
M

et
as

ta
tic

 D
is

ea
se

%
 M

et
as

ta
si

s 
Fr

ee

Hsueh	et	al.	ASCO 2016

Time	(months)

n=159
p<0.00001S

u
rv
iv
a
l	
P
ro
b
a
b
il
it
y

1.0

0.8

0.6

0.4

0.2

0

Additional single-center study8

Class 1

Class 2

p<0.00001

%
 S

ur
vi

va
l

-30

-20

-10

0

10

20

30

40

50

60

1 2 3 4 5 6

Class 1 Class 2

%
 o

f p
hy

si
ci

an
s 

re
co

m
m

en
di

ng
(c

om
pa

re
d

to
w

ith
ou

tG
E

P
re

su
lt)

Imaging SLNB

-30

-20

-10

0

10

20

30

40

50

60

1 2 3 4 5 6

*

* * *

*
* *

* *

*

*
* *

*
* * *

*

*