Clinical impact of a 31-gene expression profile test on physician recommendations 

for management of melanoma patients in a prospectively tested cohort

Martin Fleming, MD1, Clare Johnson, RN2, Kyle Covington, PhD2, Joseph Gadzia, MD3, Larry Dillon, MD4, Federico A. Monzon, MD2
1The University of Tennessee Health Science Center, Memphis, TN; 2Castle Biosciences, Inc., Friendswood, TX; 3Kansas Medical Clinic, Topeka, KS; 4Dr. Larry Dillon, Colorado Springs, CO

Background

• A 31-gene expression profile (GEP) test which identifies
cutaneous melanoma tumors as low risk (Class 1) or high risk
(Class 2) of metastasis has been clinically validated.1-4

• The test has been shown to influence physicians to direct
clinical management of cutaneous melanoma patients in
several clinical use studies (Table 1).5-7

• To further assess the clinical impact of the GEP test, we
undertook a study to evaluate and compare clinical
management plans prospectively, including initial workup,
follow-up intervals, and referral patterns, established by
physicians prior to and after GEP testing.

• Here we present preliminary results of this multicenter,
prospective clinical utility study to determine the clinical
impact of the GEP test on patient management plans.

Results

Table 2. Cohort demographics

Conclusions

• Overall, 46% of tested patients had a change in clinical management
• The majority of reported management changes were in a risk-appropriate direction, with 81% of decreases in care provided to low-

risk Class 1 patients and 87% of increases in care provided to high-risk Class 2 patients

• Physicians used GEP results to individualize management based on biological risk, as determined by the test, while still remaining
within the context of established practice guidelines

• Results of this prospective study show that the accurate identification of risk provided by the GEP informs appropriate clinical
management and patient care. The change in management is similar to three additional clinical utility studies.

References

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. J Clin Oncol 2017;34:9581.
4. Hsueh EC, et al. J Hematol Oncol 2017;10:152.
5. Berger AC, et al. Curr Med Res Opin 201632:1599-604.
6. Farberg AS, et al. J Drugs Dermatol 2017;16:428-31.
7. Schuitevoerder D, et al. Ann Surg Oncol 2017;24:S144.

Disclosures

CJ, KC and FAM are employees and stockholders of Castle Biosciences, Inc.

The proprietary GEP test is clinically available through Castle Biosciences as 
the DecisionDx®-Melanoma test (www.SkinMelanoma.com).

Table 3. Clinical and molecular features across treatment
groups

Clinical Characteristics
Overall

n=127

Median age (range), years 63 (28-95)
T stage
T1 61 (48%)
T2 32 (25%)
T3 17 (13%)
T4 8 (6%)
Not reported 9 (7%)

Breslow thickness
Median (range), mm 1.0 (0.1-18.0)
≤1 mm 66 (52%)
>1 mm 61 (48%)

Mitotic index
<1/mm2 78 (61%)
≥1/mm2 49 (31%)

Ulceration
Absent 103 (81%)
Present 20 (16%)

Growth pattern
Superficial spreading 30 (24%)
Nodular 16 (13%)
Desmoplastic 6 (5%)
Lentigo maligna 3 (2%)
Other/not assessed 72 (56%)

Site
Trunk 43 (34%)
Extremity 66 (52%)
Head and neck 18 (14%)

GEP result
Class 1 96 (76%)
Class 2 31 (24%)

Feature

Dermatology

n=41

Surgical 

Oncology

n=82

Medical 

Oncology 

n=4

Breslowa* 0.5 (0.1-4.9) 1.2 (0.0-7.5) 1.7 (0.2-18.0)
Ulcerationb

Absent 88% (36) 78% (64) 75% (3)
Present 12% (5) 17% (14) 25% (1)

Mitosisb

<1/mm2 68% (28) 60% (49) 25% (1)
≥1/mm2 32% (13) 40% (33) 75% (3)

GEP Classb

Class 1 83% (34) 71% (58) 100% (4)
Class 2 17% (7) 29% (24) 0% (0)

Class 1 Class 2

Decrease Increase Decrease Increase

Labs* 3 0 1 7

Imaging* 4 0 2 14

Adjuvant 0 0 0 1

Visits* 14 4 0 10

Referral* 13 2 3 6

aMedian (range), bPercent (count), *p<0.001

*p≤0.005, Fisher’s exact test

Figure 1. Number of cases increasing or decreasing intensity of
management by GEP Class

Table 4. Frequency of each modality of change in patients with
decreases or increases in intensity of clinical management

Figure 2. Schematic representation of risk stratification using
AJCC stage with GEP test result to guide patients’ clinical
management

Melanoma

Staging 

(per NCCN)

DecisionDx 

Low risk: 

Class 1

DecisionDx 

High risk: 

Class 2

Increase

Intensity 

(Trials)

Decrease

Intensity

Continue High 

Intensity mgmt

(Encourage trials)

DecisionDx 

Low risk: 

Class 1

DecisionDx 

High risk: 

Class 2

Low Intensity 

mgmt

(Derm)

AJCC Low risk
Stage I, IIA

AJCC High risk
Stage IIB, IIC, III

Early detection of 

recurrence

Appropriate treatment 

or clinical trial

Methods

• The RT-PCR-based GEP test was performed using primary
tumor tissue. Metastatic risk class was determined using a
proprietary predictive modeling algorithm which provides a
binary classification of Class 1 (low risk) or Class 2 (high-risk).

• At initial evaluation, prior to GEP testing, each patient’s
baseline data was assessed. Physicians’ pre-test
recommendations for follow-up were collected and
categorized as laboratory tests (labs), imaging, clinical visits,
adjuvant treatment discussion, and referral to surgical or
medical oncology.

• At the subsequent visit following receipt of GEP test result,
follow-up recommendations were again collected to capture
any changes in management.

• Changes were categorized as increases, decreases or no
change based on comparison of management plans pre-
and post-receipt of GEP test result.

Study Result

Berger (2016)5

Prospective, multicenter
n patients = 163

53% changed mgmt after 
inclusion of GEP result

Farberg (2017)6

Dermatologist survey
n physicians = 169

47-50% changed mgmt
after inclusion of GEP result

Schuitevoerder (2017)7

Prospective, single center
n patients = 91

52% of mgmt decision 
based on GEP result using 

decision tree model

Table 1. Management changes in three clinical use studies

Methods

• Of 204 patients enrolled in the study, 127 patients from 15
dermatology, medical oncology and surgical oncology
centers completed study participation at time of censoring
(June 30, 2017).

Of 36 Class 1 patients who changed, 
81% had reduced surveillance 

intensity and/or referral

Of 23 Class 2 who changed, 
86% had increased surveillance 

intensity and/or referral 

Overall 46% of patients (59/127) 
had a documented 

change in management 
following test result

Class 1: 38% (36/96) of 
patients changed

Class 2: 74% (23/31) 
of patients changed


	FC17PosterCastleBiosciencesFlemingClinicalImpact.pdf