Pathological Responses to Low-Dose Irradiation and Pepleomycin in Oral Cancer are Predictive of Locoregional Control

The prognosis of advanced oral cancer remains dismal. While multimodal therapy is benecial, maintaining the quality of life of long-term survivors is important. Therefore, risk-adapted treatment regimens need to be designed. We herein investigated whether pathological responses in oral cancer patients treated with preoperative chemoradiotherapy predict locoregional recurrence. of and pepleomycin, to and Pathological responses were dened based on the grading system of Oboshi and Shimosato.

locoregional control and overall survival (OS), and increase the potential for limited surgery on primary sites in patients with resectable locally advanced OSCC [3].
Since the treatment outcomes of head and neck cancer are improved by multidisciplinary therapy, efforts to maintain the quality of life (QOL) of long-term survivors are crucial. Adjuvant radiotherapy after surgery for head and neck cancer is associated with late adverse events, including xerostomia and dysphagia, which deteriorate QOL [4]. Lower-dose irradiation to the parotid gland and swallowing structures is associated with fewer and less severe late adverse events, resulting in better QOL [5,6]. However, less intensive therapy, such as lower-dose irradiation, may increase the risk of recurrence, and oral cancer patients with recurrence, regardless of an initially early stage, have a poor prognosis [7]. Therefore, riskadapted treatment regimens need to be designed. The di culties associated with the selection of initial treatment strategies, particularly adjuvant radiotherapy and less or more intensive treatments, are attributed to the challenges of maintaining the balance between the preservation of function and cosmesis and locoregional control in head and neck cancer. Therefore, the identi cation of prognostic factors is crucial for predicting the prognosis of each patient and selecting appropriate optimal therapy.
We performed preoperative chemoradiotherapy consisting of low-dose irradiation and pepleomycin followed by surgery for oral cancer patients with primary tumors of 3 cm or larger or clinically positive lymph node (LN) metastases in an attempt to improve curability and enable less invasive surgery. In the present study, we investigated whether pathological responses in oral cancer patients treated with preoperative chemoradiotherapy predict locoregional recurrence.

Patient eligibility
Between January 2009 and June 2018, 89 patients with oral cancer were referred to the Department of Radiation Oncology, Shizuoka City Shizuoka Hospital. Among these patients, 7 who received palliative radiotherapy, 6 curative radiotherapy without surgery for either primary or recurrent disease, 14 postoperative radiotherapy with or without preoperative radiotherapy, and 10 preoperative radiotherapy and concurrent systematic therapy other than pepleomycin were excluded. The remaining 51 patients treated with preoperative radiotherapy and concurrent pepleomycin followed by curative surgery were examined in the present study.
The pretreatment work-up included a physical examination, contrast-enhanced computed tomography (CECT) unless contraindicated and/or magnetic resonance imaging (MRI) to assess the tumor extent, and/or 2-deoxy-2[F-18] fluoro-D-glucose positron emission tomography (FDG-PET) to detect LN metastases and distant metastases. Patient staging was revised according to the 7th UICC clinical staging system. All tumors were diagnosed as squamous cell carcinoma by a histopathological examination of biopsy specimens. This retrospective study was approved by the Institutional Review Board, and written informed consent was waived because of its retrospective design.

Treatment
Radiotherapy was performed using EXL-15SP (Mitsubishi) with energy of 6MVX before July 2016 and with Clinac iX (Varian) with energy of 4 or 10 MVX after July 2016. Patients were treated with two-or three-dimensional conformal radiotherapy. Each patient received external beam irradiation to the primary tumor and lymphatics (2 Gy per day for approximately 3 weeks). The total dose was selected based on the degree of acute mucositis and extent of shrinkage of the primary tumor at the discretion of the treating physician. The elective nodal eld was mainly decided based on the primary site and extent of disease. Concurrent chemotherapy with low-dose pepleomycin (2.5 mg/day) was administered subcutaneously in a 10-hour continuous injection daily to prevent interstitial pneumonia during radiation [8]. Surgery was performed approximately 3 weeks after the completion of preoperative chemoradiotherapy. Primary tumor surgery with or without neck dissection was conducted. Limited surgery was performed on patients with a favorable clinical response in soft tissue organs, such as the tongue and buccal mucosa. Neck dissection was conducted based on the risk of nodal spread and discretion of the surgeon. Comprehensive or selective neck dissection was performed depending on the extent of the primary tumor, primary tumor thickness, number and site of clinical lymph node metastasis, and presence/absence of extranodal extension. In reconstruction, a radial forearm ap, rectus abdominis ap, scapula osteocutaneous ap, and latissimus dorsi ap were used where necessary. The safety margins of the primary lesion were 10-20 mm from palpable margins. Frozen sections obtained during surgery were examined to confirm adequate margins.

Histopathological evaluation
Pathological responses were de ned based on the grading system of Oboshi and Shimosato [9,10]. Grade 0 corresponds to no response. In grade 1 corresponding to a minor response, cancer cells are damaged without the destruction of the tumor structure. In grades 2a and 2b, destruction of the tumor structure is observed in addition to damaged cancer cells (in less than 75% of the analyzed tissue in grade 2a and more than 75% in grade 2b). In grade 3, non-viable cancer cells are present. In grade 4, no cancer cells are found. A favorable pathological response is defined as the absence of any viable tumor cells (grades 3 and 4), equivalent to a complete response. An unfavorable pathological response was defined as the presence of any viable tumor cells (grades 0-2b).

Follow-up
After the completion of surgery, patients were followed up every month by a clinical examination and every 3 months by computed tomography during the first 2 years, and every 6 months thereafter until death or data censoring.
Outcomes and statistical analysis OS and disease-speci c survival (DSS) were calculated from the date of the start of chemoradiotherapy to the date of death from any cause or from oral cancer, respectively. Time to locoregional recurrence was calculated from the date of the start of chemoradiotherapy to the date of rst local or regional recurrence, respectively.
A survival analysis was conducted using the Kaplan-Meier method and Log-rank test. Statistical analyses were performed with Fisher's exact test for categorical variables or the Wilcoxon rank-sum test for numerical variables. All tests were two-sided at a significance level of 0.05. Analyses were performed using R version 3.5.2 (The R Foundation for Statistical Computing, Vienna, Austria).

Patients
Baseline clinical and treatment characteristics are shown in Table 1     The median follow-up period was 52.1 months. At the end of the follow-up, 13 patients had disease progression and 11 had died. The cause of death was disease progression in 8 patients, sepsis due to other diseases in 2, and colon cancer in 1. The sites of failure were locoregional recurrence in 13 patients and distant recurrence in 6. All locoregional recurrence occurred as the rst site of recurrence. Three-year OS, 3-year DSS, and 3-year locoregional control rates in all patients were 78.6, 84.7, and 75.1%, respectively (Fig. 1). Besides the number of pathological LN metastases, the pathological response was identi ed as the only signi cant prognostic factor for locoregional control in the univariate analysis (Table 3). Three-year locoregional control rates were 100 and 56.6% in patients with favorable and unfavorable pathological responses, respectively. The locoregional control rate was signi cantly higher for patients with favorable pathological responses than for those with unfavorable responses (p < 0.05; Fig. 2). A favorable pathological response correlated with DSS (p < 0.05; Fig. 3) and was associated with OS (p = 0.05; Fig. 4). No other relationships were observed between pathological responses and other clinicopathological factors, including the number of pathological LN metastases or treatment-related factors (Table 4).

Discussion
In the present study, we demonstrated that pathological responses to preoperative chemoradiotherapy predicted locoregional control rates. Although our patients received low-dose irradiation (median 30 Gy, range 22-36 Gy), locoregional recurrence was not detected in those with favorable pathological responses. The present results indicate the potential of a favorable pathological response as a good surrogate marker for controlling microscopic and macroscopic tumors even with low-dose irradiation in combination with chemotherapy and surgery. Low-dose irradiation may result in fewer adverse events and better QOL.
Previous studies reported a relationship between the prognosis of patients with oral or oropharyngeal cancer and responses to preoperative chemoradiotherapy [10,11]. Patients with stage II-IV oral or oropharyngeal cancer were treated with preoperative chemoradiotherapy (50 Gy in 25 daily fractions and concurrent chemotherapy with mitomycin C and uorouracil), followed by radical surgery. Two-thirds of patients were responders, who were de ned as having no vital tumor or minimal tumor remnants encompassing less than 5%. Locoregional control was signi cantly better for responders than for nonresponders (2 years; 92.4-94.2% vs. 68.0-69.8%, p < 0.001) [11]. Kirita et al. reported a relationship between survival and pathological responses after preoperative chemoradiotherapy (40 Gy in 20 daily fractions, cisplatin-or carboplatin-based chemotherapy) in patients with stage II-IV tongue cancer. Responders were de ned as grades 2b, 3, and 4 based on the Oboshi/Shimosato classi cation. Progression-free survival rates were higher for responders than for non-responders (2 years; 95.7-100 vs. 50%). Favorable pathological responses (grades 3/4 in 25/5 patients, respectively) were observed in 30 patients (69.8%) based on our de nition [10]. The present study showed a favorable pathological response based on the Oboshi/Shimosato classi cation in 41.2% of patients, which was lower than that in previous studies. This discrepancy may be attributed to lower-dose preoperative irradiation and differences in chemotherapy regimens, pathological response criteria, and patient backgrounds. However, the prognosis of responders in the present study was similar to that in previous studies, even though our patients received lower-dose irradiation. Therefore, lower-dose irradiation may be su cient for patients with favorable pathological responses. To the best of our knowledge, few studies have investigated the relationship between the prognosis of patients with oral cancer and responses to preoperative low-dose chemoradiotherapy.
Surgical resection may exert adverse effects on appearance, swallowing, speech, and shoulder function, while the addition of adjuvant radiotherapy may cause xerostomia, altered taste, dental decay, and osteoradionecrosis with deteriorating dysphagia [12]. Combined treatments may provide the best chance of a cure at the cost of more frequent and severe adverse events and lower QOL. A systematic review of 26 retrospective studies on QOL in head and neck cancer patients treated with surgery alone or in combination with adjuvant radiotherapy demonstrated that the addition of adjuvant radiation worsened mouth dryness, thick saliva, and di culty with mouth opening [4]. The current National Comprehensive Cancer Network guidelines recommend adjuvant radiotherapy of at least 44 and 60 Gy at 2 Gy/fraction to lower and higher risk regions of the neck, respectively [1]. Radiation dose-response relationships were previously reported between swallowing structures and late dysphagia [13] and the parotid glands and late xerostomia [5,14]. Lee et al. demonstrated that the threshold value of the parotid mean dose for the incidence of LENT-SOMA grade 3 or higher xerostomia was 20 Gy [5]; however, there was no threshold dose for RTOG/EORTC grade 4 xerostomia in the study by Dijkema et al. [14]. Severe late xerostomia was observed in 50% of patients with parotid mean doses of 39.9-43.6 Gy [5,14]. A QUANTEC review recommended that the mean dose to each parotid gland needs to be kept as low as possible, consistent with desired clinical target volume coverage [15]. Levendag et al. reported that the threshold value in the superior constrictor muscle, one of the swallowing structures, for the incidence of RTOG grade 3 or more dysphagia was 21 Gy [13]. Intensity-modulated radiation therapy (IMRT) is a useful technique for delivering a high dose to the tumor and minimizing the dose to organs at risk (OAR) by varying the beam intensity in the radiation eld. This technique was previously shown to reduce the incidence and degree of late adverse events, including late xerostomia and dysphagia, without decreasing locoregional tumor control [16,17]. However, in a situation in which parotid glands and swallowing structures are within or near tumor tissues, IMRT cannot reduce the radiation dose for these OAR su ciently when standard highdose radiotherapy consisting of 70 Gy in curative settings or 60 to 66 Gy in adjuvant settings was delivered in 2-Gy fractions. Previous studies demonstrated that the average mean doses to pharyngeal constrictors and the ipsilateral/contralateral parotid gland were 58 and 47.6/25.4 Gy, respectively, in head and neck cancer patients treated with standard curative radiotherapy, even with the IMRT technique [16,17]. Therefore, lower-dose irradiation consisting of 30 Gy in 2-Gy fractions in the present study is considered to offer the clinically meaningful bene t of alleviating late radiation toxicity. Ongoing clinical trials are investigating whether dose de-escalations may be safely performed without the worsening of treatment outcomes in head and neck cancer patients, mainly human papilloma virus (HPV)-related oropharyngeal cancer patients, in curative or adjuvant radiotherapy settings [18]. Although the prognosis of HPV-related head and neck cancer patients is good [19], the low incidence of HPV-related oral cancer (13.8% by p16 positivity) suggests a minor role in oncogenesis in Japanese oral cancer patients [20].
Although the majority of our patients were considered to have HPV-unrelated oral cancer, preoperative lower-dose radiotherapy with chemotherapy and surgery may be su cient for tumor control in patients with favorable pathological responses.
The Japanese oral cancer guidelines recommend pre-or postoperative chemoradiotherapy for better locoregional control and OS in patients with advanced oral cancer [3]. The NCCN guidelines recommend postoperative radiotherapy for oral cancer patients with adverse risk features, including extranodal extension, a positive margin, pT3 or T4 primary tumors, N2 or N3 nodal disease, nodal disease at level IV or V, perineural invasion, vascular embolism, and lymphatic invasion, without describing their recommendations on preoperative radiotherapy for oral cancer. Since our patients were treated with preoperative chemoradiotherapy and surgery, the pathological data obtained cannot be compared with those from patients undergoing upfront surgery. Systematic reviews on the diagnostic accuracy of oral cancer demonstrated that CT and MRI were useful for evaluating the extension of the primary tumor site, and FDG-PET may contribute to the detection of metastatic cervical lymph nodes [21]. Clinical staging by multi-modality imaging is consistent with pathological staging in oral cancer. The pretreatment work-up in the present study revealed that 76.5% of our patients had clinical T3-4 and/or clinical N2-3 and/or recurrent disease, while the remaining 12 patients had clinical T2N0-1. Occult metastases have been reported in 13-33% of T1 tumors and 37-53% of T2 tumors at the time of diagnosis, even in clinical early T-stage and node-negative oral cancer patients classi ed by morphological imaging only [22]. Some of our patients with clinical T2N0-1 may have lymph node metastases pathologically and be suitable for postoperative radiotherapy. Therefore, the majority of our patients receiving preoperative chemoradiotherapy and undergoing surgery may be candidates for upfront surgery and postoperative chemoradiotherapy.
The present study has several limitations. The sample size in this single-institution study was small.
Furthermore, it was a retrospective study that may have had a selection bias. Several patients with locoregionally advanced disease underwent other treatments, including curative radiotherapy or upfront surgery with postoperative radiotherapy because they were older or had comorbidities. Therefore, the present results cannot be generalized to locoregionally advanced oral cancer patients with an older age and comorbidities. Furthermore, we selected pepleomycin, a bleomycin derivative with less pulmonary toxicity than bleomycin, as the radiation sensitizer because it was expected to be less myelotoxic than platinum-based chemotherapy even though it has been associated with an increased risk of pulmonary brosis and more severe mucositis. Concurrent chemoradiotherapy with bleomycin has been administered to patients with head and neck cancer and esophageal cancer [23,24]. Platinum-based chemoradiotherapy is now the standard treatment regimen for these cancers, partly because of the pulmonary toxicity of bleomycin and clinical evidence supporting the utility of platinum-based chemoradiotherapy in various cancers. None of the patients in the present study developed pulmonary brosis, which may have been due to the continuous injection of pepleomycin [8]. Another limitation is that the adverse events of radiotherapy, including dry mouth and dysphagia, were not systematically reported for patients in the present study, although low-dose irradiation was expected to alleviate toxicity.
Preclinical studies previously demonstrated that the administration of pepleomycin sensitized the effects of radiation on cancer cells, but not on normal oral mucosal or salivary gland cells [25,26]. The tumor speci city of this drug may enable lower-dose irradiation to control oral cancer with negligible damage to normal tissues. Further studies are needed to con rm these results. In addition, surgical and radiation techniques were non-uniform because they were selected at the discretion of the surgeon and radiation oncologist. However, no signi cant differences were observed in surgical or radiation techniques between patients with favorable and unfavorable pathological responses.

Conclusions
The present results demonstrated that pathological tumor responses to preoperative chemoradiotherapy are useful predictive factors for locoregional control and DSS. Low-dose irradiation may be su cient for patients with good pathological responses; therefore, these patients may avoid radiation-induced late toxicity associated with higher doses of irradiation. On the other hand, patients with unfavorable pathological responses need to receive more aggressive treatments, such as more intensive chemotherapy, higher doses of preoperative irradiation, or the addition of adjuvant irradiation. Overall survival rates in patients with favorable versus unfavorable pathological responses