PowerPoint Presentation


Patients with T1–T2 tumors from previously published multicenter cohort studies who

had undergone the SLNB procedure were analyzed by both the i31-GEP and the

MSKCC nomogram (n=465).6 Accuracy metrics were compared using <5% predicted

risk as a negative result and ≥5% as a positive result.

The i31-GEP for sentinel lymph node (SLN) biopsy outperforms the MSKCC nomogram in predicting the risk
of having a positive SLN in patients with cutaneous melanoma

Background
› National Comprehensive Cancer Network (NCCN) guidelines recommend

foregoing sentinel lymph node biopsy (SLNB) if the population-based point-

estimate risk of positivity is <5% (T1a with no high-risk features), discuss and

consider SLNB if the risk is 5-10% (T1a with high-risk feature(s), T1b), and

recommend SLNB if the risk is >10% (T2-T4).1

› With the current SLNB positivity rate at approximately 12%,2,3 better tools are
needed to refine patient selection for the procedure to avoid potential complications

and additional healthcare costs. Such methods that improve patient selection of

those who will have a positive SLNB, while identifying the correct patients with low

enough risk (<5% positivity risk by current guidelines) that they can safely forego

SLNB could reduce the number of unnecessary surgical procedures, lower

healthcare costs, and improve patient care.

› Two tools, the i31-gene expression profile test for SLNB (i31-GEP for SLNB)4 and
the nomogram developed at the Memorial Sloan Kettering Cancer Center

(MSKCC)5,6 predict the risk of SLN positivity in patients with cutaneous melanoma

(CM) by combining a tumor’s molecular risk profile, clinical and pathological factors

(i31-GEP test) clinical and pathological factors only (MSKCC).

› We compared the performance of the i31-GEP for SLNB result prediction to that of
the MSKCC nomogram.

Acknowledgments & Disclosures
References

› Funding provided by Castle Biosciences.
› DZ and NB have no conflicts of interest. SKM is an employee and stock/options holder at Castle Biosciences.

Methods

Presented at the 2023 Winter Clinical-Miami Dermatology Conference.

Danny Zakria, MD1, Sonia K. Morgan-Linnell, PhD2, Nicholas Brownstone, MD3

1Department of Dermatology, Icahn School of Medicine at Mount Sinai, New York, NY; 2Castle Biosciences, Inc., Friendswood, TX; 3Department of Dermatology, Temple Health, Philadelphia, PA

› The MSKCC nomogram using clinical and pathological factors
alone had a 10% miss rate in patients it predicted to have

<5% risk of SLN positivity—worse than AJCC staging alone.

› The i31-GEP for SLNB missed 2.7%, significantly lower than
MSKCC—better than both MSKCC and AJCC staging.

› The i31-GEP for SLNB showed an 89% increase in the
number of patients who could forego SLNB compared with

current guidelines (36:1 vs. 19:1 true to false negative ratio)

compared with a 53% decrease if using MSKCC.

› The i31-GEP for SLNB has demonstrated clinical utility to
guide SLNB decisions in patients with T1-T2 tumors as well as

guiding subsequent treatment plans with risk-of-recurrence.

Conclusions

Scan or click here  
for more info

1. National Comprehensive Cancer Guidelines, v2, 2022.  2. Joyce KM, et al. Ir J Med Sci 186:847–853, 2017. 3. Chen et al. 
Oncotarget 7:45671-45677. 4. Whitman ED, et al. JCO Precision Oncology 1466–1479, 2021. 5. Wong SL, et al. Ann Surg Oncol 
12:282–288, 2005. 6. Pasquali S, et al. Eur J Surg Oncol 37:675–80, 2011. 

Figure 1. i31-GEP SLNB guidance results in fewer missed 
positive SLNs than the MSKCC nomogram or standard 

staging guidance

› The i31-GEP had a significantly lower false negative rate (3/111; 2.7%) than 
the MSKCC nomogram (11/110; 10.0%) (p=0.026).

› The i31-GEP also had a lower false negative rate than using AJCC staging, 
which does not recommend SLNB if the risk is 5% or less.

Table 1: Variables included in i31-GEP test or MSKCC Model

Prediction Variables Included in

i31-GEP Test

Included in

MSKCC Model

31-GEP continuous score √

Breslow thickness √ √

Mitotic Rate √

Ulceration √ √

Age √ √

Clark Level √

Tumor location √

Table 3. Reclassification of risk in patients with 5-10% risk 
(NCCN/AJCC T1b tumors) for whom guidance is not definitive

Test
Predicted 
<5% risk

Positivity rate 
in <5% group

Predicted 
>10% risk

Positivity rate 
in >10% group

Total 
reclassified

i31-GEP
36.2% 

(46/127)
2.2%
(1/46)

15.7% 
(20/127)

15%
(3/20)

52.0% 
(66/127)

MSKCC
29.9%

(38/127)
7.9%
(3/38)

2.4%
(3/127)

0%
(0/3)

32.3% 
(41/127)

All results are % (n/N).
i31-GEP MSKCCAJCC

Table 2. The i31-GEP for SLNB has higher accuracy than MSKCC for 
predicting SLN positivity in patients with T1-T2 tumors

Test Sensitivity Specificity
Negative 
predictive 

value

Positive 
predictive 

value

i31-GEP 94.8% 26.5% 97.3% 15.5%

MSKCC 81.0% 24.3% 90.0% 13.2%

19:1 9:136:1

Missed positive SLNTrue negative SLN (safely forego SLNB)

31-GEP: Better than 
standard of care

MSKCC: Worse than 
standard of care

1 missed positive SLN per 
37 considered low-risk

1 missed positive SLN 
per 20 considered 

low-risk

1 missed positive SLN per 
10 considered low-risk

› Using only clinical and pathologic features to predict SLN status
limits the ability to identify tumors with spread to the SLN (e.g., 12%

positivity rate). Integrating clinical and pathological factors with

molecular tumor biology assessed by the prospectively validated

gene expression risk profile (31-GEP) improves risk stratification to

guide personalized patient care.

› Beyond the utility for SLNB guidance, the integrated 31-GEP ( i31-
GEP) also provides a precise risk for recurrence, metastasis, and

disease-specific mortality for an individual patient.

Clinical Impact

31-GEP score, Breslow thickness, mitotic rate, age, and Clark level were continuous variables. Ulceration was present or absent, and tumor location

was entered as trunk, extremity, or head/neck. Tumor location was evaluated in development of the i31-GEP but was not significant for prediction;

Clark level was not evaluated because most providers use Breslow thickness instead.

https://castlebiosciences.com/research-development/publications/

	Slide 1