Basal and one-month differed neutrophil, lymphocyte and platelet values and their ratios strongly predict the efficacy of checkpoint inhibitors immunotherapy in patients with advanced BRAF wild-type melanoma

The medical treatment of metastatic melanoma (MM) has significantly improved thanks to the identification of specific genetic alterations which became important therapeutic targets for new treatments as the combination of anti-BRAF + anti-MEK drugs for patients harbouring BRAF V600 mutations [1], and thanks to the new immunological approaches with monoclonal antibodies directed against the immune checkpoint molecules CTLA-4 (cytotoxic T lymphocyte antigen-4) and PD-1 (programmed death antigen 1) or its ligand PD-L-1. The anti-CTLA-4 ipilimumab is able to induce a response rate of approximately 15% with about 20% of patients remaining long-term responders [2]. The anti-PD-1 nivolumab and pembrolizumab, alone or in combination with ipilimumab, provide higher response rates of approximately 40% and almost 60%, respectively, with the majority of responses being durable [3, 4]. Despite these results, more than a half of patients do not respond or relapse early to checkpoint inhibitors and their prognosis is very poor.

Several potential biomarkers have been studied in the effort to select responder patients from non-responders with the aim to avoid useless high-cost therapies and potentially severe adverse events in non-responders. Unfortunately, no validated biomarkers are currently available in melanoma.

Serum lactate dehydrogenase (LDH) is the first prognostic blood biomarker included in the American Joint Committee on Cancer (AJCC) staging system for patients with MM. LDH is an essential enzyme for melanoma cells metabolism catalyzing the reversible conversion of pyruvate into lactate. Also, the accumulated lactate was proved to promote tumor immune escape by reducing the survival and cytolytic capacity of CD8 + T cells and natural killer cells. LDH resulted a strong prognostic marker independent of type of treatment. The prognostic value of LDH was confirmed in several pivotal and real-life studies using ICI (1–4).

Among more recent biomarkers, only higher levels of PD-L-1 expression seems correlated with a better response, but their impact on overall survival is still questionable [5].

In recent years, there has been an increasing interest in establishing the role of basal white blood cells (WBC), neutrophil, lymphocyte and platelet count and the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) as prognostic and predictive biomarkers to immune checkpoint inhibitors (ICI). These findings derive from the known contribution of neutrophils and platelets to cancer cell growth and migration, and from the evidence of the inter-relationships between thrombosis, inflammation and cancer immune surveillance [6‐11]. Beyond their crucial role in hemostasis and thrombosis, platelets are increasingly recognized as regulators of inflammation, angiogenesis and immunomodulation in cancer progression. Furthermore, platelets cover the cancer cells, inducing their protection from the attack of the immune cells. In this way, in addition to enhancing the metastatic power of the tumor, they increase the risk of formation of microthrombi and embolic events. Through the modulation of the immune system and the interaction with leukocytes, platelets influence inflammatory processes in cancer at various stages: by altering the activation state of the endothelium, recruiting leukocytes in tumor sites and tuning the inflammatory environment in sites of primary and metastatic tumors. Finally, platelets modulate innate leukocyte effector functions such as antigen presentation by dendritic cells, monocyte recruitment and differentiation or neutrophil extracellular trap formation. Finally, platelets and their microvesicles have further been shown to inhibit activation of T-helper cell and natural killer cell functions [8‐11].

High baseline levels of NLR and PLR alone or combined have been associated with poor survival and high risk of recurrence regardless of cancer type [12, 13]. In addition, they appear to predict worse efficacy both of chemotherapy and immunotherapy suggesting their potential role as agnostic markers of efficacy [14‐16].

Regarding melanoma, the prognostic role of pre-treatment NLR on PFS and OS has been reported in a large meta-analysis of patients at different stages of disease, showing a negative role independently of the disease stage and the type of treatments used [17]. The prognostic value of pre-treatment NLR in the era of ICI therapy has been evaluated in small subsets of patients affected by different type of neoplasms. However, the utility of NLR in the context of the modern immunotherapy has not been well-defined. In metastatic disease higher level of basal neutrophil and NLR resulted associated with poorer clinical outcomes in patients receiving ipilimumab [18‐20] or anti-PD-1 treatment [21‐23].

The prognostic and predictive role of platelets in melanoma has been less investigated and the modest data published so far are controversial [15, 24, 25].

Finally, it is still debated if early changes in hematological parameters during ICI therapy could increase the magnitude in discriminating responder from non-responder patients. In this regard, very few and conflicting data are available yet [15, 16, 25‐27].

Our study aimed to evaluate neutrophil, lymphocyte and platelet values and their ratios both at baseline and after one course of ICI therapy in order to verify if their variations could increase their prognostic and predictive value in patients with MM treated with ICI.

Patients with the BRAF mutation were excluded from our analysis because most of them were treated as first line with anti-BRAF/antiMEK targeted therapy. We thought that this pre-treatment would have confounding effects on clinical outcomes leading to a strong bias of patient selection (only progressing patients would have been included in the study analysis).

ICI therapy is the backbone of MM treatment, especially for patients without BRAF mutation for whom the anti-BRAF/anti-MEK target therapy is not suitable. The anti-PD-1 agents pembrolizumab and nivolumab give about 40% of response rate, with a PFS of 7–9 months and an OS of about 33–37 months [3, 4, 30].

In our retrospective cohort of patients, we reported an ORR of 41.9% with a DCR of 56.6%. The median PFS was 10 months, and the median OS was 29 months. This OS resulted a little shorter than that reported in large, controlled studies. The reasons could be that our patient population was quite different from that of controlled studies. First of all, we excluded patients harboring BRAF V600 mutations normally treated with targeted therapy as first line. Moreover, we had 10% of patients with brain metastases and 12% of patients with mucosal or uveal melanoma, that are notoriously poorly responsive to ICI therapy. Finally, 20.9% of our patients were treated with ipilimumab alone as first line, which gives a median OS of about 16–20 months [4.30]. When we recalculated PFS and OS by excluding patients with worse prognosis (brain metastases, mucosal melanoma, uveal melanoma), we reported a median PFS of 12 [7–19] months and a median OS of 42 [27–61] months, which are slightly better than those of pivotal studies.

Interestingly, increasing evidence are demonstrating that systemic and chronic inflammation mediated by different blood cell types and humoral factors induces an immunosuppressive status in the microenvironment that favors the growth and migration of cancer cells [31]. In this context, neutrophils and platelets play a relevant role [6‐10, 32]. Neutrophils have been reported to support the development of metastases through multiple mechanisms such as the release of proteases that degrade antitumoral factors, the production of leukotrienes that propagate metastasis-initiating cells, the inhibition of antitumor T cell responses [11, 33]. In contrast, tumor growth is inhibited when migration of neutrophils to tumoral areas or granulocyte colony stimulating factors are blocked [34]. Neutrophils and NLR have been reported as prognostic markers in several types of neoplasms [35] as renal cell carcinoma [32], non-small cell lung cancer [36], breast cancer [37], colorectal cancer [38].

Also, platelets and PLR have been found to be independent prognostic factors for numerous solid tumors [39] including non-small cell lung cancer [36], hepatocellular carcinoma [12], renal cell carcinoma [32], breast cancer [13], colorectal cancer [40] and melanoma [24].

Recent findings demonstrated that also the response to ICI is influenced by the inflammation status. Higher levels of neutrophils have been found associated with less benefit from ICI treatment in different types of cancer [14‐16, 19‐23]. Moreover, Zhou and colleagues used a whole-blood multicolor flow cytometry in patients with metastatic cancers to identify a liquid immune profile signature based on five cell subtypes significantly associated with a better OS: CD14^high monocytes, CD8 + /PD-1⁺ T cells, plasmacytoid dendritic cells, neutrophils, and CD3⁺/CD56⁺/CD16⁺ natural killer T cells. Of note, in the time point after the first administration of ICI, only natural killer T cells and neutrophils still maintained their predictive value for OS and PFS, underlining the relevant role played by neutrophils in the responsiveness to ICI immunotherapy [41].

NLR and PLR were also reported as strong prognostic markers in patients with non-small cell lung cancer treated with nivolumab [42]. Furthermore, a pan-immune-inflammation value (PIV) involving the neutrophil, platelet and monocyte count was significantly correlated with worse clinical outcomes in 163 microsatellite instability-high metastatic colorectal cancer patients treated with ICI [38].

In MM, it has been reported that basal neutrophils and NLR are prognostic in patients receiving ipilimumab [43] or anti-PD-1 [21, 23]. These data have been recently confirmed in a pooled from MM patients accrued in three large phase II/III randomized ICI trials, one of which included a chemotherapy arm [44]. Higher values of NLR, interleukin-6 and C reactive protein were associated with a shorter survival in patients receiving ICI or chemotherapy [44]. Finally, in a retrospective study including 72 patients with MM and 84 with non-small-cell lung cancer treated with PD-1 inhibitors, Kartolo and colleagues reported that a baseline value of NLR ≥ 5 and PLR ≥ 200 were independently associated with worse ICI efficacy regardless of cancer type. In addition, patients with both high NLR and PLR had the worst treatment outcome when compared to patients with only one elevated biomarker or none [15].

Of interest, some recent reports underlined that early changes of these simple hematological parameters during ICI therapy increase the capacity to distinguish responder from non-responder patients. However, very few data are currently available on this topic in particular regarding early variations of platelets and PLR.

In 197 patients with advanced melanoma treated with ipilimumab, Cassidy and colleagues reported that an increase of 30% in the NLR during treatment was strongly associated with worse OS. Interestingly, this change was not prognostic in the 65 patients treated with BRAF inhibitors leading authors to conclude that NLR may have a uniquely predictive value in patients treated with ICI [26]. Another study in patients with renal cell carcinoma treated with ICI demonstrated that a ≥ 25% increase in NLR resulted strongly associated with a worse PFS and OS [27]. More recently, in 224 patients with advanced melanoma treated with PD-1 inhibitor monotherapy, while confirming the independent predictive value of baseline NLR on survival and its correlation with performance status and number of metastatic sites, Bartlett and colleagues also found that an increase in NLR ≥ 30% during the first 2 cycles of treatment was associated with a significantly worse OS and with a trend towards shorter time to treatment failure. Interestingly, combining baseline NLR ≥ 5 and a NLR increase ≥ 30%, the Authors identified a small cohort of patients with a very poor OS [16]. In a small group of 16 MM patients treated with anti-PD1, Ohashi and colleagues found that responders presented an increase of lymphocytes and eosinophils and a decrease of neutrophils within the first 6 weeks of treatment [25]. Finally, Corti and colleagues reported that an early increase of pan-immune-inflammation value (PIV) involving the neutrophil, platelet and monocyte count, resulted an independent predictor of clinical benefit in metastatic colorectal cancer patients treated with immune checkpoint inhibitors [38].

Nevertheless, although confirming the predictive role of 6-week NLR and PLR, other Authors were not able to establish an optimal cut-off for ΔNLR and ΔPLR in prognosticating treatment efficacy, probably due to the fact that the majority of patients maintained high values of NLR and PLR at 6-week of ICI therapy [15].

In our paper, we showed that variations between time 1 and time 0 (Δ) of platelets, WBC, neutrophils, and lymphocytes strongly influenced the PFS in a two-variable modelMoreover, patients with ΔNLR superior to the cut-off had a worse PFS (HR = 1.975, p < 0.001), as well as patients with ΔPLR over the cut-off (HR = 1.693, p = 0.001). In our experience, all these parameters have also significantly affected the OS and response.

Nevertheless, in the multivariable model only ΔNLR resulted statistically significant on PFS (p = 0.004) with an increase above the cut-off causing a 71% higher risk of progression. Interestingly, when we clustered our patient population in 4 groups according to basal LDH value and the ΔNLR, we found that LDH ≤ ULN and ΔNLR under the cut-off identified a sub-group of patients with the longest median PFS of 20 months versus only 5 months of patients with both these parameters elevated. Of interest, patients with normal LDH but with ΔNLR above the cut-off also had a median PFS of about 5 months showing that ΔNLR had the strongest impact on PFS outcome.

The OS, in multivariable analysis, resulted influenced by WBC, lymphocytes and PLR variations with ΔWBC and Δlymphocytes having a statistically significant: the increase above the WBC cut-off causes a 92% greater probability of death (p < 0.001) and the decrease of lymphocytes greater than the cut-off is associated with a worse OS (HR = 1.627, p = 0.038).

When we clustered the entire patient population in 4 categories according to basal LDH and Δlymphocytes, we found that LDH ≤ ULN and a higher Δlymphocytes identified patients with the longest median OS of 50 months versus only 10 months of patients with LDH > ULN and a lower Δlymphocytes. Also, combining ΔWBC and Δlymphocytes, we found that ΔWBC lower and Δlymphocytes higher identified patients with a longer OS of 45 months versus 7 months of patients with ΔWBC higher and Δlymphocytes lower.

Finally, in the multivariable model, the DCR was associated with changes in WBC and lymphocytes, but only the increase of WBC above the cut-off value resulted significantly associated with a poorer DCR (OR = 3.422, p = 0.003).

Multiple NLR and PLR cut-offs were suggested by previous studies for different cancer types including melanoma [15, 16, 21, 45]. For clinical practice purposes, it would be convenient for clinicians to have unifying NLR and PLR cut-off values regardless of cancer types.

We used a two-variables regression model for all haematological parameters considering their baseline values as continuous variables and the one-month differences as categorical. Then, the difference between time 1 and time 0 of WBC, neutrophils, lymphocytes, platelets and their ratios (NLR and PLR) were calculated and dichotomized using cut-off values assessed by ROC curve. For the analysis of the one-month differences in haematological parameters, two-variable regression was used to take into account for possible confounding influences of their baseline values.

The novelty and strength of our work compared to the others previously published include: homogeneity and large size of our population with all patient being BRAF wild type and treated with checkpoint inhibitors as first line; evaluation of all parameters studied both at baseline and after one course of ICI therapy; addition of the role of platelets and PLR, hitherto little studied, as predictors of response to ICI. All these simple and cost-effective hematological parameters, both at basal and as early variations, resulted to predict the efficacy of treatment in a large population of MM.

Due to the facility to obtain these parameters through routine procedures, they could be used as potential biomarkers in decisions about candidacy allowing clinicians to choose patients most likely to have a clinical benefit to ICI therapy, and for stratification in clinical trials. Moreover, their early variations during treatment may serve as a simple monitoring tool to evaluated the continuation of therapy in the case of uncertain clinical outcomes.

There were some limitations in our study that require attention. First of all, the retrospective nature of our study. Moreover, the heterogeneity of our patient’s population (10% with brain metastases, 12% with mucosal or uveal melanoma, 20.9% treated with ipilimumab alone) and the exclusion of BRAF V600 mutated patients. Lastly, in the absence of standardized values that define cut-off for these hematological parameters and for their ratios, we also used arbitrary cut-offs obtained from the ROC curves of OS. To overcome these limitations, further evaluations in larger multicenter prospective studies are warranted.