Transcriptomic Approach Provides Molecular Classification of PTCL Subtypes

Take-aways:

• A refined classification algorithm and standardized assay distinguished the transcriptomic signatures of PTCL subtypes defined by the World Health Organization.
• Results from formalin-fixed, paraffin-embedded tissue samples were frequently concordant with results from fresh-frozen tissue samples, allowing wider clinical applications than previous efforts to molecularly classify PTCL.
• The comprehensive molecular classification of PTCL subtypes can have clinical relevance as a potential diagnostic tool.

---

A new transcriptomic approach to the molecular classification of peripheral T-cell lymphoma (PTCL) subtypes may have potential applications as a diagnostic tool.

Catalina Amador, MD, of the University of Nebraska Medical Center, and colleagues developed the approach because PTCL, a form of non-Hodgkin lymphoma, “includes heterogeneous clinicopathologic entities with numerous diagnostic and treatment challenges.” For example, the World Health Organization (WHO) recognizes more than 25 subtypes of PTCL, yet 30% of PTCL cases do not belong to any of the specific subtypes.

Dr. Amador and colleagues developed a refined classification algorithm and standardized assay based on previous research in which they “defined robust transcriptomic signatures that distinguish common PTCL entities and identified two novel biologic and prognostic PTCL-not otherwise specified (NOS) subtypes” using fresh-frozen samples with transcriptome-wide arrays.

However, this practice had limited applications in routine clinical use because it used fresh-frozen samples. While formalin-fixed, paraffin-embedded samples are more commonly used in routine diagnosis, they can present obstacles to obtaining diagnostic signatures, as formalin fixation can cause DNA and RNA to become fragmented, cross-linked, or chemically modified.

“Therefore, the effective translation of our highly accurate RNA-based PTCL diagnostic signatures to [formalin-fixed, paraffin-embedded] tissue is challenging but essential to implement an assay with wide clinical application,” Dr. Amador and colleagues wrote.

They aimed to consolidate the PTCL diagnostic signatures from fresh-frozen RNA samples into a single technical platform that can determine PTCL subtype diagnoses using formalin-fixed, paraffin-embedded tissue samples.

Study Design and Methodology

Dr. Amador and colleagues constructed a training cohort comprised of samples from 105 patients with PTCL who had previously generated gene expression profiling data and matched fresh-frozen and formalin-fixed paraffin-embedded tissue samples. The researchers also assembled a validation cohort of 140 cases that had not been previously analyzed. The two cohorts had similar clinical outcomes and comparable distributions of PTCL subtypes.

PTCL cases were centrally reviewed and diagnosed per the current WHO PTCL subtype classifications. Three hematopathologists “thoroughly reevaluated” cases in the validation cohort and used a comprehensive immunostaining panel and T-cell receptor gene rearrangement analysis when needed, according to the researchers.

They used data from the training cohort to select classifier genes for the nCounter platform and selected training cohort tissue samples that had adequate tumor tissue and RNA quality and assessed transcriptomic signatures on HG-U133plus2.0 and nCounter. HG-U133plus2.0 was used for fresh-frozen samples, while nCounter was used for formalin-fixed paraffin-embedded samples.

Outcomes

The transcriptomic signature assessment results from the nCounter and HG-U133plus2.0 platforms were highly correlated in approximately 60% of signature-specific genes. The investigators used a recursive filtering analysis to generate 11 to 20 diagnostic transcripts for each subtype of PTCL and 16 housekeeping genes. These “well-performing” transcripts did not impact the classification accuracy, sensitivity, or specificity in fresh-frozen samples analyzed by HG-U133plus2.0 or formalin-fixed, paraffin-embedded samples analyzed by nCounter, according to the researchers.

The molecular subclassification using formalin-fixed paraffin-embedded samples was “highly comparable” with the “gold standard” of fresh-frozen samples and had an error rate of less than 5% across PTCL subtypes, with “highly concordant results” when two additional laboratory sites performed the assay, the study’s authors wrote.

The refined transcriptomic classifier in formalin-fixed paraffin-embedded tissue samples from the training cohort had sensitivity of >80%, specificity of >95%, and accuracy of >94% compared with the fresh-frozen tissue samples. See TABLE 1 for data on different subgroups.

In the validation cohort, the transcriptional classifier data generated from formalin-fixed paraffin-embedded samples matched the diagnosis from expert hematopathologists in 85% of cases, while there was a “borderline association” with the molecular signatures in 6%, and a disagreement between the transcriptional classifier and experts in 8%, according to the study’s authors.

The classifier “improved the pathology diagnosis” in two cases, while four of the disagreement cases had a “molecular classification that may provide an improvement over pathology diagnosis on the basis of overall transcriptomic and morphological features,” Dr. Amador and colleagues wrote.

The transcriptional classifier separated two novel molecular subgroups of PTCL-NOS, PTCL-GATA3 and PTCL-TBX21, with an 87% concordance in the training cohort. In the validation cohort, the nCounter platform classified 40% of PTCL-NOS cases as PTCL-GATA3 and 52% as PTCL-TBX21.

The researchers also analyzed 12 cases of PTCL-T follicular helper that they excluded from the validation cohort. They found two cases had a significant association with the AITL molecular signature, while four cases had “borderline scores” between AITL and PTCL-NOS, and six cases had a “clear association” with PTCL-NOS. Of those six cases, three resembled PTCL-TBX31, and three resembled PTCL-GATA3. However, PTCL-T follicular helper is “unlikely to be a single entity as currently defined and needs further evaluation to determine how best it should be characterized,” the study’s authors noted.

Limitations and Conclusions

The study’s limitations included small sample sizes for some PTCL subtypes, as well as the lack of gene expression profile data available for the validation cohort.

“We believe that it can facilitate a better definition of cases for research studies and ensure more uniform cohorts for clinical trials,” Dr. Amador and colleagues concluded.

The research was supported, in part, by grants from Allos Therapeutics and Spectrum Pharmaceuticals.

Reference

Amador C, Bouska A, Wright G, et al. Gene expression signatures for the accurate diagnosis of peripheral T-cell lymphoma entities in the routine clinical practice. J Clin Oncol. 2022. doi:10.1200/JCO.21.02707