Mature tertiary lymphoid structures (mTLS) — B-cell follicles with CD23+ follicular dendritic cells — predict response to immune checkpoint inhibitors across many cancers, sometimes outperforming PD-L1 and tumor mutational burden. But their promise depends on being able to score them reliably in an ordinary pathology lab, and criteria have varied study to study. Across 357 carcinoma and sarcoma samples, this study built and validated a routine-technique workflow — HES staining plus CD20 and CD23 immunohistochemistry — that detects mTLS as sensitively as the multiplex immunofluorescence used in research, with near-perfect agreement between pathologists on maturity. The result is a portable, standardized method ready for clinical trial use.
This validation study in Laboratory Investigation, led by Dr. Lucile Vanhersecke and Dr. François Le Loarer at Institut Bergonié with the Cordeliers immunology group of Prof. Wolf H. Fridman and Prof. Catherine Sautès-Fridman and Prof. Antoine Italiano, set out to make mature tertiary lymphoid structure (mTLS) scoring reproducible in routine pathology. Building on the team’s earlier retrospective and prospective PEMBROSARC-linked work showing mTLS predict checkpoint-inhibitor benefit, it benchmarks a conventional-stain workflow against the multiplex immunofluorescence used in the original research setting. Explicyte performed the blinded multiplex immunofluorescence analysis that served as the reference standard against which the routine-pathology method was validated.
The routine-pathology workflow was validated against multiplex immunofluorescence as the benchmark. Explicyte analyzed the multiplex-IF assays blind (with W.H. Fridman), scoring CD20/CD21/CD23 whole-slide images in HALO software to identify follicular dendritic cell networks and call TLS maturity. This mIF read-out was the reference against which HES and CD20/CD23 immunohistochemistry were measured across the cohort.
In 62 TLS-positive samples scored blind by both methods, all 30 mTLS called by IHC were confirmed by multiplex IF, and 40 of 42 (95.2%) mIF-mature cases were confirmed by IHC. Where double CD20/CD23 staining missed follicular dendritic cells (11 of 40 cases, 27.5%), adding a single CD23 stain rescued the maturity call in 10 of 11 (90.9%) — establishing single CD20 plus single CD23 as the most sensitive routine combination, with HES-visible germinal centers needing no immunostaining at all.
Reviewing 240 samples from 97 TLS-positive patients, the team showed detection depends heavily on the specimen. Surgical resections were 6.1× more likely to contain TLS than biopsies (95% CI 3.5–10.7; P < 0.0001), and primary tumors about 2× more likely than metastases. TLS also clustered at the invasive tumor front rather than the tumor core — practical guidance for where and what to sample.
Four pathologists (two trainees, two seniors) scored a 67-sample set after brief training. Interrater agreement was substantial for TLS presence (Fleiss κ 0.65) and near-perfect for maturity (Fleiss κ 0.90). Residual disagreements centered on interpreting isolated CD23+ cells, prompting the authors to publish a shared “atlas of TLS” to standardize these gray-zone calls.
The study closes the gap between a research biomarker and a clinical one: mTLS can now be scored the same way, sensitively and reproducibly, in any pathology lab with standard stains.
For trial sponsors and pathology labs, this provides a validated, low-cost path to use mTLS status as a patient-selection biomarker without specialized multiplex platforms — increasingly relevant as mTLS enters immunotherapy trials as an inclusion criterion (e.g., NCT04095208, NCT04874311, NCT04968106). The sampling findings matter operationally: biopsies and metastatic samples under-detect TLS, so trial design and reporting should standardize specimen type, and scoring should default to IHC when crush artifact degrades HES. A shared training atlas is the mechanism for keeping scoring consistent across sites.
