The approval of the chemo-immunotherapy combinations and targeted therapies for resectable non-small cell lung cancer (NSCLC) marked the beginning of a new therapeutic paradigm in early-stage lung cancer management. However, with advances come challenges, and as neoadjuvant immunotherapy gains traction in thoracic oncology, assessing treatment response remains a challenge.
Chee Lee, MD, Associate Professor, University of Sydney, and the co-chair of the 2023 World Conference on Lung Cancer, set the stage for a discussion on this subject during the meeting’s closing plenary session, saying, “with neoadjuvant treatment emerging as the standard of care in routine practice, the work of our pathology colleagues in assessing treatment response has major implications for clinical care of our patients.”
The session, Advances in Lung Cancer Pathology, took place September 12 and is available on demand for registered WCLC 2023 attendees through December 31.
The Role of Pathology in the Neoadjuvant Setting
Keith M. Kerr, BSc, MB, ChB, Professor of Pathology, Aberdeen University Medical School, UK, spoke first on the role of pathology in the neoadjuvant setting.
He outlined the three key uses of pathology in the this setting:
- Pre-operative diagnosis and staging:
- Identifying patients for biomarker-guided therapy; and
- Pathologic assessment of response (as a surrogate for overall survival) in surgically resected specimens after neoadjuvant therapy.
Definitive surgery is curative in a significant proportion of patients with early-stage NSCLC. Therefore, a careful risk-benefit analysis is needed before adding neoadjuvant therapy, Dr. Kerr said.
While the prognostic impact of conventional TNM features is well known, histopathologic features of the resected tumor may also provide prognostic information and inform patient selection for neoadjuvant therapy. For example, Dr. Kerr referred to a study published in the February 2023 issue of Lung Cancer showing that high-grade growth patterns in lung adenocarcinoma biopsies predict poor survival after surgery.
Biomarker-guided neoadjuvant therapy is being investigated in ongoing trials, including the LCMC4 LEADER study. These studies have practice-changing potential for selecting patients who may benefit from neoadjuvant targeted therapy.
Immunotherapy and chemo-immunotherapy combinations in the neoadjuvant setting demonstrate efficacy in most studies, in terms of response rates and pathologic response assessment, Dr. Kerr said. However, the evidence for biomarkers predictive of response to immunotherapy/chemoimmunotherapy in the neoadjuvant setting is not as clear cut.
Nevertheless, there is a role for pathology in the neoadjuvant chemoimmunotherapy setting. In the CheckMate 816 study of neoadjuvant nivolumab plus platinum-based chemotherapy in patients with resectable NSCLC, event-free survival (EFS) correlated with PD-L1 expression levels. Although EFS benefit was higher in patients with higher PD-L1 expression, patients with PD-L1 expression <1% also obtained some benefit.
Based on these data, the US Food and Drug Administration (FDA) approved neoadjuvant nivolumab plus platinum-based chemotherapy for patients with early-stage NSCLC lacking actionable alterations in EGFR and ALK, regardless of their PD-L1 status. Dr. Kerr noted that pathologists would be involved in determining the EGFR and ALK status in this context, even if PD-L1 assessment is not a prerequisite.
The European Medicine Agency (EMA) has approved the same neoadjuvant combination for resectable NSCLC, but only in patients with PD-L1 expression ≥1%. The use of this regimen in the European Union would require a pathologist to assess EGFR, ALK, and PD-L1 status.
Dr. Kerr noted that, unlike other cancers, assessment of major pathologic response (MPR) and pathologic complete response (pCR) is not common in lung cancer practice because data-driven validated recommendations are currently lacking.
Closing the Gap in Assessing Pathologic Response
IASLC Pathology Committee Chair Sanja Dacic, MD, PhD, Vice Chair and Director of Anatomic Pathology, Yale School of Medicine, New Haven, Connectictu, presented data from the IASCLC Interobserver study. Speaking about the motivation for the study, she said, “The IASLC leadership recognized that there is a huge practice gap in the assessment of pathologic response.”
To address this gap, the IASLC Pathology Committee initiated the MPR Project and subsequently published multidisciplinary recommendations for standardized processing and pathologic assessment of surgical specimens following neoadjuvant therapy.
“With all these recommendations in the background, the next question was how good are pathologists at assessing response and how reproducible are their assessments,” she added.
The next phase of the MPR project focused on evaluating the reproducibility of the IASCLC histologic criteria (published in 2020) in a larger dataset of 772 slides of resected tumor specimens from 84 patients. These patients had received neoadjuvant treatment in one of these six trials—LCMC3, NADIM, NEOSTAR, MPDL3280A, TOP 1501, or MED14736. The slides were evaluated by 11 pathologists, including five members of the IASLC Pathology Committee and six researchers involved in the clinical trials. All pathologists underwent training to facilitate understanding and application of the ISALC criteria and MPR calculator tool.
Pathologic response was calculated as a percentage, based on the ratio of the “viable tumor” area to the total tumor area. The MPR calculator tool estimated MPR/pCR status based on unweighted (average tumor bed area) and weighted (normalized to the tumor bed area on each slide) percentage of the viable tumor.
The data showed 81% unanimous agreement between pathologists. There were no differences between weighted and unweighted MPR assessments, suggesting that the simpler unweighted approach can be used for routine practice.
Concordance between pathologists was highest at the two extremes, i.e., inter-observer agreement (IRA) was 0.98 in cases with >95% viable tumor and 0.94 in cases with 0% viable tumor. Dr. Dacic spoke to the significance of these data, noting that 0% viable tumor in the resected specimen may serve as a surrogate for pCR in the lymph nodes, even though lymph nodes were not included in this study, and that, “we can reassure you that pathologists can do an excellent job in assessing pCR.”
Potential contributors to inter-observer discordance include difficulties in determining tumor bed; visual separation of viable tumor bed from admixed stromal inflammation; differentiating between reactive pneumocytes and lepidic adenocarcinoma; mucinous adenocarcinoma interpretation; and possible data entry errors.
Dr. Dacic pointed out some outstanding issues with pathologic response assessment—the role of immunohistochemistry and other ancillary techniques; lymph node evaluation and the approach to integrating the results; and the use of digital pathology and artificial intelligence (AI). She mentioned that the ISALC Pathology Committee has initiated studies focused on the latter two issues.
She discussed recent data from the ISALC Pathology Committee demonstrating concordance between AI-powered assessment of pathologic response and visual MPR in differentiating disease-free survival (DFS) and OS rates, using resection specimens after neoadjuvant atezolizumab in the LCMC3 trial cohort.
“You have to keep in mind that in order for any AI tool to be implemented in a clinical trial, and particularly, in routine practice, it has to be validated and re-validated,” she said.
In the next phase of the MPR project, an updated version of the AI algorithm will be applied to the same dataset, minus the LCMC3 cohort, to calculate the percentage of viable tumor and classify MPR status.
In her summary, Dr. Dacic noted that the IASLC MPR study confirms that pathologists agree, for the most part, on pathologic response assessments. She also noted that pathologists showed excellent reliability in cases with no viable tumor and good reliability for MPR assessment with the IASLC recommended cut-off of ≤10%.