Background

Head and neck squamous cell carcinoma (HNSCC) accounts for nearly 7% of the global cancer burden [1], and approximately 66% of HNSCC patients are diagnosed at an advanced stage [2].

RADPLAT, a combination of radiotherapy and trans-arterial chemotherapy, is widely reported as an effective treatment [3]. Moreover, patients can maintain their basic functions, including speech, mastication, and swallowing by RADPLAT. However, not all patients can obtain the expected outcomes; some patients may develop resistance to treatment and may experience relapse after a complete response. This challenge is not only for RADPLAT but also for any other treatment for HNSCC, which remains a significant clinical concern. With regard to the prognosis of head and neck cancers, one report indicated that the 5-year survival rate is estimated to be 50% globally [4]. Another report stated that it varies from approximately 46.8% to 71.4% [5]—with this variability being influenced by various factors, including tumor characteristics and patient-related variables. Although some factors related to treatment outcome, including T-stage classification or tumor volume, have been reported, these factors cannot predict which patient could achieve prolonged survival. Thus far, we have not reached a consensus on which biomarkers are appropriate predictors for treatment outcomes.

Recently, some reports have indicated that absolute lymphocyte count (ALC) is related to a prognostic indicator in particular cancers [6]. With the development of immunotherapies, the role of ALC as a parameter reflecting immune reactions has been highlighted. Most reports investigating the relationship between ALC and prognosis have suggested that low ALC levels are associated with poor prognosis although the underlying mechanisms have been explored [6]. However, the field of head and neck cancer has not reached a consensus on ALC as a predictive biomarker, and finding a specific role for ALC in relation to prognosis may lead to potentially improved outcomes for all HNSCC.

Furthermore, although the current evidence level supporting RADPLAT remains low, identifying ALC as a predictor for the treatment outcome of RADPLAT could contribute to the establishment of more precise, personalized treatment strategies, which may lead to an increase in the patients receiving RADPLAT. The broader application of RADPLAT in HNSCC patients may lead to improvements in the post-treatment quality of life. Thus, we consider well-defined criteria for screening favorable candidates for RADPLAT are required to establish it as a standard treatment option.

Patient Selection

Patients over 20 years old diagnosed with oral cancers and who underwent RADPLAT from May 2011 to February 2023 were enrolled. The exclusion criteria included a history of treatment for the same primary, having concurrent treatments like systemic chemotherapy or molecularly targeted drugs. All patients were histologically diagnosed with squamous cell carcinoma. We reviewed the lab records (i.e. weekly blood cell counts) of all patients to examine the association between ALC and recurrence in each case.

Treatment

In RADPLAT, the number of cisplatin infusions was 1-6 cycles, and 40 - 66 Gy of external irradiation was given to the primary and metastatic lymph nodes. Our intra-arterial chemotherapy was different from the original RADPLAT in intervals between treatments. We repeated intra-arterial chemotherapy every 3- 4 weeks. Irradiation was performed with 3D conformal radiation therapy (3D-CRT) or intensity-modulated radiation therapy (IMRT). The irradiation dose depends on the treatment purpose, whether curative or preoperative. Patients received 66 Gy of irradiation for curative treatment, and others received 40 Gy as preoperative treatment. Irradiation was delivered in conventional fractionations.

Statistical analysis

We constructed progression-free survival (PFS) curves using the Kaplan-Meier method, and used the log-rank test to compare the survival curves between ALC at each point. We also assessed the survival curves based on the difference in ALC between two arbitrary points and its association between ALC recovery and recurrence outcomes using the same method.

Receiver operating characteristic (ROC) curves were constructed for ALC based on the relapse outcome. P values were judged to be significantly different when P < 0.05. Statistical analyses were performed using Python v3.10.11 and R v4.3.0.