In a study reported in The Lancet Oncology, Brian Rini, MD, of the Cleveland Clinic Taussig Cancer Institute, and colleagues developed a 16-gene assay and recurrence score that predicted postoperative outcome in patients with stage I to III clear cell renal cell carcinoma.¹

Development Phase

In the development phase, examination of the association between expression of 732 genes measured by reverse transcription–polymerase chain reaction and clinical outcome in 942 patients with stage I to III clear cell renal cell carcinoma who had undergone nephrectomy at the Cleveland Clinic yielded 516 genes that were associated with a recurrence-free interval.

In this population, 63% of patients were male; the median age was 62 years (26% ≥ 70 years); 58% had radical surgery; 49% had a tumor size ≤ 4 cm; stage was I in 68%, II in 10%, and III in 22%; necrosis was present in 17%; Fuhrman grade was 1 in 0.4%, 2 in 56%, 3 in 35%, and 4 in 9%; 0.5% had positive nodes; and 23% had recurrence.

Of the 516 genes, 11 were selected by additional statistical analyses and were combined with 5 reference genes, with a recurrence score (0–100) algorithm being developed. The genes included in the assay consisted of those involved in vascular processes (APOLD1, EDNRB, NOS3, PPAP2B), cell growth/division (EIF4EBP1, TUBB2A, LMNB1), immune response (CEACAM1, CX3CL1, CCL5), and inflammation (IL-6) as well as the reference genes (AAMP, ARF1, ATP5E, GPX1, RPLP1).

Validation Phase

In the validation phase, the recurrence score was investigated in a French cohort of 626 patients. In this population, 71% of patients were male (P = .002 vs development cohort); the median age was 61 years (29% ≥ 70 years); 64% had radical surgery (P = .01); 46% had a tumor size ≤ 4 cm; stage was I in 64%, II in 9%, and III in 28% (P = .03); necrosis was present in 33% (P < .0001); Fuhrman grade was 1 in 4%, 2 in 30%, 3 in 46%, and 4 in 20% (P < .0001); 0.6% had positive nodes; and 16% had recurrence.

Risk Prediction

On univariate analysis, the continuous recurrence score (median = 37) was associated with a recurrence-free interval, with a hazard ratio (HR) of 3.91 (P < .0001) for each 25-unit increase in the score. Other significant predictors on univariate analysis were tumor size (HR = 2.63, P = .002, for ≤ 4 vs > 4 cm), Fuhrman grade (HR = 2.84, P < .0001, for 3–4 vs 1–2), and Leibovich score (HRs = 4.31 for high vs low and 3.08 for intermediate vs low, P = .01).

On multivariate analysis, the recurrence score was a significant predictor of recurrence, with a hazard ratio of 3.37 (P < .0001) for each 25-unit increase after stratification by stage and adjustment for tumor size and Fuhrman grade. No significant interaction of recurrence score with any of the covariates for recurrence-free interval was observed (all P > .10).

In multivariate analysis including the Leibovich score, the recurrence score remained a significant predictor of recurrence, with a hazard ratio of 4.20 (P < .0001) per 25-unit increase, and the Leibovich score was no longer a significant predictor (P = .06). Tumor size remained a significant predictor on multivariate analysis (HR = 2.09, P = .02). The C statistic for recurrence with recurrence score alone was 0.79; the addition of recurrence score to the Leibovich score improved the C statistic to 0.81 from 0.74 with the Leibovich score alone.

Change in Risk Grouping

The recurrence score identified patients with stage I disease at high risk and patients with stage II to III disease at low risk. The recurrence score thresholds of 32 and 44 were identified post hoc as distinguishing risk levels, with scores < 32 indicating low risk, scores 32–44 indicating intermediate risk, and scores > 44 indicating high risk.

On this basis, 39% of 398 stage I patients were at low risk, with a mean 5-year recurrence risk of 2%, and 15% were at high risk, with a mean 5-year recurrence risk of 23%. Among patients with stage II to III disease, 19% of 224 were at low risk, with a mean 5-year recurrence risk of 2%, and 44% were at high risk, with a mean 5-year recurrence risk of 39%. The recurrence score risk groupings distinguished 5-year overall survival and renal cancer–specific survival rates in patients with stage I disease and in those with stage II to III disease.

The investigators concluded: “Our findings validate the recurrence score as a predictor of clinical outcome in patients with stage I to III clear cell renal cell carcinoma, providing a more accurate and individualized risk assessment beyond existing clinical and pathological parameters.” ■

Disclosure: The study was funded by Genomic Health Inc and Pfizer Inc. For full disclosures of the study authors, visit www.thelancet.com.

Reference

1. Rini B, Goddard A, Knezevic D, et al: A 16-gene assay to predict recurrence after surgery in localised renal cell carcinoma: Development and validation studies. Lancet Oncol 16:676-685, 2015.