Brisk Tumor-Infiltrating Lymphocytes May Hold Prognostic Value in Primary Cutaneous Melanoma

Tumor-infiltrating lymphocytes (TILs) are a known and important biomarker for monitoring immune therapy response in patients with melanoma, but research is lacking regarding their potential function as a prognostic factor. A large study published in JAMA Network Open suggests that brisk TILs could be an independent prognostic factor for overall survival (OS) in melanoma.

TILs are reliable as a biomarker, but past research has produced varied results on their value in prognosis for cutaneous melanoma, which is the most aggressive form of common skin cancer. TILs are widely classified as absent, present, nonbrisk (infiltrating focally, not along the entire base of the tumor), and brisk (infiltrating the entire tumor or entire base of the tumor).

One barrier to large-cohort studies on TILs in melanoma is that manual review and aggregation of medical records is a labor-intensive undertaking. The current study aimed to gauge the prognostic significance of TILs in the largest cohort to date of patients with cutaneous melanoma that was established using natural language processing (NLP) algorithms.

Researchers identified patients with primary cutaneous melanoma in the Brigham and Women’s Hospital pathology laboratory information system between June 1, 2004, and December 31, 2019. They used NLP to identify patients whose records included primary invasive melanomas with TIL grades assessed and vertical growth phase documentation. Then, data on 7 characteristics were extracted and normalized into groups. These factor were including TIL grade, Breslow thickness, mitotic rate, ulceration, histologic regression, microsatellites, and vascular or lymphatic invasion.

The pathology reports identified by the NLP were linked to patients’ electronic health record data and to Massachusetts death certificate files from 2004 to 2019, allowing researchers to gather patient demographic information and death information to calculate OS.

In total, 14,436 patients with cutaneous melanoma were identified, and a cohort of 2624 patients had vertical growth phase melanoma with TIL status scored. The cohort was 55.7% male and the median age was 61 years, with an interquartile range of 50 to 72 years. Median follow-up time was 3.1 years.

Of those patients, 434 (16.5%) had disease classified as absent TILs, 1916 patients (73%) had nonbrisk TILs, and 274 patients (10.4%) had brisk TILs. For patients with absent TILs, the 5-year survival rate was 71.0% (95% CI, 65.5%-76.9%). Patients with nonbrisk TILs had a 73.8% (95% CI, 71.1%-76.5%) 5-year survival rate, and those with brisk TILs had an 85.2% survival rate (95% CI, 80.0%-90.7%). In the overall cohort, the 5-year survival rate was 74.3% (95% CI, 72.1%-76.5%).

Brisk TILs were identified as an independent prognostic factor associated with improved OS (adjusted HR, 0.63; 95% CI, 0.42-0.95; P = .03). Those patients had a 14.2% OS improvement at the 5-year mark compared with those classified as having absent TILs. Nonbrisk TILs were not associated with an OS improvement compared with the absence of TILs (adjusted HR, 0.87; 95% CI, 0.68-1.11; P = .25).

Overall, patients who were younger, were female, and had lower Breslow thickness, lower mitotic rate, brisk TILs, absence of ulceration, absence of microscopic satellites, and absence of vascular or lymphatic invasion had improved OS, which is consistent with previous studies.

One study takeaway was the effectiveness of utilizing NLP to establish large cohorts with detailed histopathologic features for survival analysis. In regard to TILs, the data provide evidence that brisk TILs can be a prognostic factor for OS in primary cutaneous melanoma.

The authors concluded, “Based on our findings, we suggest that TIL grade be included in future AJCC staging revisions and routinely incorporated in a standardized manner into primary cutaneous melanoma pathology reports.”

Reference

Yang J, Lian J, Chin YP, et al. Assessing the prognostic significance of tumor-infiltrating lymphocytes in patients with melanoma using pathologic features identified by natural language processing. JAMA Netw Open. 2021;4(9):e2126337. doi:10.1001/jamanetworkopen.2021.26337