Prognostic values of hematological biomarkers in nasopharyngeal carcinoma patients treated with intensity-modulated radiotherapy

Table 1 lists the general clinical characteristics of 427 enrolled patients with 307 (71.9%) males and 120 (28.1%) females. The median age was 48 years (range: 17–82 years). There were 213 (49.9%) patients with locally advanced diseases (T3–4), 383 (89.7%) patients with nodal metastasis of the neck. In terms of tumor-node-metastasis (TNM) staging, there were 9 (2.1%) patients in stage I, 80 (18.7%) in stage II, 208 (48.7%) in stage III and 130 (30.5%) in stage IV. All patients completed the planned course of treatment with 59 (13.8%) patients who received radiotherapy alone and 368 (86.2%) patients who received concurrent chemoradiotherapy with or without neoadjuvant/adjuvant chemotherapy.

During a median follow-up of 67.5 months (range 4.8–85.5 months), there were 57 (13.3%) patients having local–regional recurrence, 64 (15.0%) patients experiencing distant metastasis, and 64 (15.0%) dead. The 5-year PFS and OS were 76.0% (median: 85.1 months, 95% CI 74.1–96.1 months) and 85.8% (mean: 77.8 months, 95% CI 76.0–79.6 months), respectively.

Table 2 presents the hematological biomarkers level at pre-treatment and post-treatment. Paired sample t test with Bonferroni correction revealed that ANC, APC, and ALC were significantly declined after therapy (P < 0.001 each), while NLR and PLR were elevated significantly (P < 0.001 each).

Since the levels of hematological biomarkers at post-treatment were influenced by various factors, such as chemotherapy and nutritional support, therefore, we only analyzed the prognostic value of pre-treatment variables on survival.

Patients with NLR greater than or equal to 2.32 (NLR ≥ 2.32) were significantly inferior compared with those less than 2.32 in respect of 5-year OS (81.8 vs 90.0%, P = 0.015) and 5-year PFS (70.9 vs 81.5%, P = 0.005). The 5-year OS for those with PLR greater than or equal to 123.0 (PLR ≥ 123.0) was not reached for those whose PLR was less than 123.0 (81.9 vs 89.9%, P = 0.011). However, a significant difference of PLR subgroups was not observed with regard to 5-year PFS (72.8 vs 79.5%, P = 0.095; Fig. 1). Compared to patients with ANC lower than 3.9, those whose ANC greater than or equal to 3.9 (ANC ≥ 3.9) had worse 5-year PFS (72.7 vs 79.8%, P = 0.030), yet it was not associated with 5-year OS (83.2 vs 88.6%, P = 0.224; Fig. 2). APC subgroups (≥ 206 vs < 206) presented insignificant differences in either 5-year OS (84.5% vs 87.1%, P = 0.766) or PFS (74.9 vs 77.2%, P = 0.581). Furthermore, ALC subgroups (≥ 1.6 vs < 1.6) showed no obvious differences in OS (86.4 vs 85.1%, P = 0.487) or PFS (76.8 vs 75.2%, P = 0.700; Fig. 2).

Unadjusted Cox regression analysis showed that high NLR (≥ 2.32) and PLR (≥ 123.0) were significantly associated with inferior OS (P = 0.017, P = 0.012 each; Table 3); high ANC (≥ 3.9) and NLR (≥ 2.32) had significant reduction in terms of PFS (P = 0.032, P = 0.005 each; Table 4). Statistical differences were not observed in ANC, APC, or ALC subgroups with regard to OS (P = 0.225, P = 0.766, P = 0.488 each; Table 3), as well as APC, ALC, or PLR subgroups in PFS (P = 0.581, P = 0.700, P = 0.096 each; Table 4).

After adjustment for some potential confounders, including age, TNM staging, and treatment modality, high NLR (≥ 2.32) was still significantly inferior in OS (HR 1.699, 95% CI 1.005–2.873, P = 0.048; Table 3) and PFS (HR 1.710, 95% CI 1.150–2.543, P = 0.008; Table 4). Moreover, high PLR (≥ 123.0) remained significantly related to worse OS (HR 1.765, 95% CI 1.051–2.964, P = 0.032; Table 3), yet it was not correlated with PFS (HR 1.318, 95% CI 0.896–1.939, P = 0.161; Table 4). Although high ANC (≥ 3.9) was insignificant in case of PFS (HR 1.461, 95% CI 0.990–2.155, P = 0.056; Table 4), its clinical value remained worth noting.

As a highly radio-sensitive disease, radiotherapy remains the first-line treatment for non-disseminated NPC. The development of IMRT is a breakthrough in the treatment of NPC. Compared with two-dimensional radiotherapy (2D-RT) and three-dimensional conformal radiotherapy (3D-CRT), IMRT generates higher radiation dosage to tumor volume with better target coverage and normal tissue sparing. IMRT has been proven with ideal local control in NPC [10], nevertheless, distant control remains inadequate. Individual therapy selection should be established for optimal prognosis. Besides treatment techniques, the biological variability of tumors cannot be overlooked any more.

Clear differences were found between levels of hematological markers tested at pre- and post-treatment in our present study. ANC, APC and ALC level showed significant decrease, while NLR and PLR increased significantly at post-treatment. Decreased ANC, APC and ALC level were most likely due to malnutrition as an impact of poor oral intake and increased protein catabolism [11, 12]. Both clinical conditions can lead into subsequent immunosuppression. On the other side, increased NLR and PLR were most likely because of systemic inflammation and critical lymphopenia caused by the treatment administered.

The relationship between inflammation and cancer has been reported to be interactive and synergetic. The sites of chronic inflammation often have greater risk of neoplasia [13, 14]. Elevated level of inflammatory cells and cytokines infiltration were frequently present in tumor biopsies [13]. Inflammatory factors in situ facilitate angiogenesis of tumor, inhibition of adaptive anti-tumor immunity, as well as elicit insensitivity of hormones regulation. Moreover, neoplasms further release cytokines and chemokines into the systemic circulation to regulate the level of lymphocytes, neutrophils and platelets [15, 16].

Lymphocytes, as important components of immune surveillance, induce tumor rejection by specific recognition of tumor-related antigens (TAs) [17]. It has been reported that a higher occurrence of neoplasms was observed in mice with a defect of lymphocytes [18], or some key components of T cell effectors, such as perforin [19], interferon-γ (IFN-γ) [20]. Further study revealed tumor regression due to intensified response of T lymphocytes, implemented through administration of interleukin 2 (IL-2) or autograft of tumor-infiltrating lymphocytes (TILs) [21, 22]. Nowadays, immune checkpoint inhibitors have achieved notable success in a variety of cancers, which potentiate lymphocyte responses by specifically effecting on cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) or programmed cell death protein 1 (PD-1) [23].

Neutrophils are major indicators of inflammation and infection. It is modulated by various cytokines, of which, interleukin-8 (IL-8) [24], macrophage inflammatory protein-1β (MIP-1β) [25], as well as macrophage migration inhibitory factor (MIF) [26] have been widely studied to implicate in development and progression of tumors. Meanwhile, neutrophils can further produce kinds of cytokines to facilitate carcinogenesis [27]. The leukemoid reaction in NPC was regarded as the primary manifestation of malignancy or relapse [28]. Peripheral neutrophils from HNSCC patients presented with a longer life span [7]. The long-lived and activated neutrophils contributed to tumor cell metastasis [29]. Peng et al. [30] further verified that tumor cells were less available to migrate with inhibited neutrophil infiltration, and the ability restored by re-infiltration. Furthermore, in vitro study reported that the responses of cytotoxic T lymphocytes were inhibited by infiltrating neutrophils, which effect was directly proportional to the amount of neutrophils [31].

Platelet, a regulatory factor in thrombosis and hemostasis, contributes a lot in tumor growth, extravasation and dissemination [32]. It is well acknowledged that hypercoagulable state is frequently presented in cancer patients [33]. The activated platelets shield tumor cells from immune eliminations, and escort them to distant sites driven by platelet-derived transforming growth factors (TGF-β) [34]. Labelle et al. [34] also demonstrated that platelet inactivation with ablation of nuclear factor kappa B (NF-κB) pathways in tumor cells or TGF-β1 could suppress the metastatic potential of tumor cells. Other study [35] revealed that platelet-derived CXCL5 and CXCL7 chemokines sped up the development of prometastatic microenvironments. Moreover, either blockade of CXCL5/7 receptor CXCR2 or depletion of platelets would slow down the metastatic process.

Our study presented NLR and PLR were independent prognosticators for survival in NPC patients receiving IMRT-based therapy. High NLR was significantly associated with inferior OS and PFS, and high PLR was correlated with poor OS. Yet, ANC, APC or ANC were insignificant. There has been a mass of studies elaborating the significant relation of ANC, APC, ALC, NLR or PLR with prognosis in different types of cancers, as well as HNSCC [1, 2, 8, 9]. However, the cutoff values were discordant. Su et al. [1] illustrated that initial ANC > 8 × 10⁹/L was correlated with poor OS, PFS and DMFS, and APC > 10 × 10⁹/L indicated inferior OS and PFS in NPC patients. He et al. [2] once divided initial peripheral lymphocytes, neutrophil and NLR from NPC patients into quarters, patients with highest either lymphocyte percentage or ALC had superior PFS compared with those of lowest quartiles. While highest neutrophil percentage or NLR > 2.74 predicted poor PFS. Sun et al. [9] reported that NPC patients with NLR ≥ 2.7 or PLR ≥ 167.2 indicated shorter PFS, while PLR ≥ 163.4 was correlated with poor OS. Lu A et al. [36] revealed that NLR ≥ 2.28 indicated poor OS and PFS, with PLR ≥ 174 correlated with inferior OS. Lin et al. [37] believed the combination of platelet with NLR (COP-NLR score) was a better predictor compared with NLR alone or APC alone. A latest meta-analysis by Su et al. [38] evaluated the hematological parameters in NPC patients, and suggested the significant role of NLR and lymphocyte in prognostic prediction. As a matter of fact, it is hard to pinpoint a unified value since each cancer has its own specific bio-physiological characteristics. In addition, the baseline levels of biomarkers were diverse. On the other hand, even for the same type of cancer, the differences in measurement exist between different instruments.

There were several advantages in our study. Though the relationship of hematological biomarkers with survival has been studied profoundly in HNSSC, we specifically focused on NPC. Since IMRT technique has been the mainstream, it is noteworthy that we carried out this study with both large-population sample size and IMRT-based therapeutic approach. Moreover, we took multiple hematological variables into consideration.

The inadequacy of our study was the lack of other inflammatory indicators, such as C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) and lactate dehydrogenase (LDH) for more comprehensive analysis. Further randomized trials of large-scale, multi-center studies are needed to elucidate the clinical values of hematological markers on prognosis and therapy plan decision.