Immuno-oncology (IO) agents have revolutionized the treatment of cancers. Targeting the regulatory immune checkpoints, inhibitors of programmed cell death 1 (PD1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA4) signaling have demonstrated efficacy and have been approved as standard therapeutic options in several cancer types [1]. In the management of recurrent or metastatic HNSCC (R/M HNSCC), pembrolizumab and nivolumab, both anti-PD1 antibodies, are now used as standard second line agents [2].

Anti-PD1 antibodies block the interaction of PD1 on CD8 T cells with the immune-suppressing ligand PD-L1 on antigen-presenting cells and tumor cells, thereby reversing immune tolerance [3]. Since the mechanism of anti-PD1 antibodies is thought to be dependent on the activity of functional T lymphocytes, it is rational to hypothesize that the efficacy of anti-PD1 antibodies would be compromised in patients with lymphopenia. To our knowledge, however, the impact of lymphocyte counts has not previously been investigated in a cohort of HNSCC patients on anti-PD1 therapies.

Importantly, growing evidence has demonstrated that conventional cancer treatment modalities may lead to significant lymphopenia [4–8]. Moreover, lymphopenia has long been associated with poorer prognosis in multiple cancer types [9–13]. This phenomenon is further highlighted in treatment regimens that involve protracted or fractionated radiation protocols such as pancreatic cancers [4], lung cancers [5], high grade gliomas [6], and HNSCCs [7]. The vast majority of patients with R/M HNSCC undergo a protracted course of chemoradiation leading to treatment-related lymphopenia, which is long-lasting. It is, however, unclear to date whether absolute lymphocyte counts (ALC) at the time of anti-PD1 therapy is correlated with responses to therapy in HNSCC patients. Similarly, increased neutrophil-to-lymphocyte ratio (NLR) has also been established as a poor prognostic factor in multiple cancer types including head and neck cancers [10, 14, 15]. While the ratio incorporates the neutrophil count in addition to the lymphocyte count, it is unclear whether the ratio provides any insight distinct from the lymphocyte count alone in the context of immunotherapy response.

In this single center retrospective review, we sought to determine whether baseline ALC prior to immunotherapy could be associated with immunotherapy response in R/M HNSCC. In parallel, we also explored whether pretreatment NLR provides additional data over ALC with regard to clinical outcomes.

Previously, we observed that CRT leads to development of significant lymphopenia in HNSCC patients [7]. In this study, we confirmed again that CRT indeed leads to lymphopenia in HNSCC patients. We further expanded this observation to show that CRT leads not only to lymphopenia but also overall leukopenia, neutropenia, decreased percent of lymphocytes, and increased NLR. We have also previously shown that lymphopenia is closely linked with prior radiation but not necessarily prior systemic chemotherapy [5]. Consistent with this, we observed in this HNSCC cohort that duration from last radiation treatment was a significant factor associated with pre-IO lymphopenia. In contrast, the number of prior systemic lines of therapy did not significantly associate with pre-IO lymphopenia although this lack of significance may be explained by the heterogeneity of the systemic regimens in terms of drugs, cumulative dose, and timing.

Importantly, low ALC was significantly associated with lack of clinical benefit from IO therapy; R/M HNSCC patients with low ALC < 600 cells/μl were associated with significantly worse response to therapy and shorter PFS. Patients in the lower ALC group had median PFS of 2 months, which mirrored those of previous studies [17–19]. Interestingly, in the higher ALC group, the median PFS was approximately 5 months. Overall, PD-1 inhibitor therapy in this study appeared to demonstrate superior ORR compared to previous studies [17–19].

To our knowledge, this study for the first time demonstrates a significant association between pretreatment lymphocyte counts and IO therapy response in a cohort of R/M HNSCC patients. Our observations provide additional evidence that the peripheral lymphocyte count may be an important parameter to consider when administering anti-PD1 therapy. These findings are particularly critical for three fundamental reasons: (i) the vast majority of locally advanced HNSCC patients will develop lymphopenia due to the first-line CRT, (ii) immunotherapy has now become standard second-line therapeutic option for patients with R/M HNSCC, and (iii) we currently do not have a way to predict which patients will respond to IO therapy. Predictive biomarkers of PD1 inhibitor therapy other than PD-L1 expression, which is of limited utility in HNSCCs, are much in need, and this study gives further support to the potential utility of ALC as such biomarkers. Mechanistically, even though the total lymphocyte count alone is far from fully representing the state of antitumor immune response, one can speculate that it may be a surrogate indicator of lymphocyte presence in the tumor microenvironment and/or the degree of immunosurveillance.

In this study, we had the opportunity to evaluate whether the incorporation of neutrophil counts into a ratio against lymphocyte counts provided any unique insight into immunosuppression in this population that might have been missed if just ALC was evaluated. In fact, the use of NLR did not provide any additional insight. Although studies have suggested the importance of tumor-associated neutrophils in pro-tumorigenic processes [20], any potential association between the absolute neutrophil counts and response to anti-PD1 therapy remains largely uncharacterized and of unclear importance. Furthermore, neutrophil counts recover more readily from prior RT compared to lymphocyte counts and may be drastically influenced by various clinical contexts, e.g. active infections or acute glucocorticoid use, confounding the utility of NLR. Additional studies are necessary to determine whether ALC is the more reliable correlate for anti-PD1 therapy response.

In cancer types other than HNSCC, prior studies have shown that pretreatment lymphocyte counts as well as NLR are potential biomarkers associated with response to IO therapy [21, 22]. Several groups have observed that ALC and/or NLR are associated with response to IO therapy in melanoma patients [14, 23–25]. We have also previously reported a significant association between these parameters and IO therapy response in a heavily heterogeneous cohort of multiple cancer types, e.g. lung, melanoma, renal cell, urothelial, and HNSCC (eight of 167 patients) [26]. In HNSCC patients, there has been a recent report showing a significant association between high NLR (> 8.77) and poorer clinical outcomes with anti-PD1 therapy [27].

Limitations of the study include both the retrospective and single-center nature of the study as well as the small size of the cohort. Also, the current cohort represented a relatively more treatment-naïve group of patients as suggested by the apparently higher ORR and longer overall PFS compared to previous studies [17–19]. Another limitation may be the inclusion of some patients with nasopharyngeal carcinoma (NPC), which was not included in landmark trials that have led to the approval of anti-PD1 antibodies in HNSCC; the use of pembrolizumab in NPCs is currently a category 2B recommendation based on the latest update of the National Comprehensive Cancer Network guidelines [2]. Additionally, it cannot be determined from our data whether lymphopenia is a truly predictive biomarker of anti-PD1 immunotherapy, or simply a global prognostic biomarker that reflects an advanced disease state irrespective of treatment intervention. Lastly, other potential prognostic markers of IO response, e.g. PDL1 expression and tumor mutational burden for non-virally associated tumors, could not be factored into the multivariate analyses.

A critical next step to confirming the utility of these correlates as predictive biomarkers of IO therapy response would be to confirm the observations in a prospective trial. It would also be valuable to establish the biologic significance and the mechanism behind peripheral lymphocyte counts and IO therapy response, e.g. whether peripheral lymphocyte counts truly correlate with the compositional changes in the tumor microenvironment and/or functional changes in the antitumor immunity, specifically in the context of CRT-associated lymphopenia. To venture further beyond these ideas, a challenging task would be to determine whether allowing for treatment-related lymphopenia to recover, when done in appropriate clinical contexts, in fact improves the response to IO therapy in the patients who actually recover. Thus, it would be interesting to explore whether treatment strategies aimed at reducing treatment/radiation-related lymphopenia in HNSCC patients would truly lead to superior clinical outcomes.

In conclusion, this study demonstrates for the first time that pretreatment ALC in R/M HNSCC patients is significantly associated with response to PD1 inhibitor therapy and offers a potential predictive biomarker to be explored further.