Neoadjuvant chemo-free combination of camrelizumab and apatinib for locally advanced resectable oral squamous cell carcinoma – A pilot study


 Novel neoadjuvant therapy regimens are needed to improve the outcomes of patients with locally advanced resectable oral squamous cell carcinoma (OSCC). We conducted a prospective, open-label, single-arm trial (n = 21, NCT04393506) to determine the safety and feasibility of neoadjuvant camrelizumab (an anti-PD-1 antibody) plus apatinib (a VEGFR inhibitor) for locally advanced resectable OSCC. The primary endpoints were safety and major pathological response (MPR). Neoadjuvant camrelizumab plus apatinib was well-tolerated and the MPR rate was 40% (8/20), meeting the primary endpoint. All five patients with CPS ˃ 10 had MPR. Additionally, patients achieving MPR showed more CD4+ T cell infiltration and a higher CD8+/FoxP3+ ratio than those without MPR (p < 0.05), and decreased CD31 and ɑ-SMA expression were observed after neoadjuvant therapy. Our findings demonstrate that neoadjuvant therapy with a chemo-free combination of camrelizumab and apatinib is safe and yields a promising MPR rate, supporting further trials.


Background
For patients with locally advanced resectable oral squamous cell carcinoma (OSCC), surgery with adjuvant radiotherapy or chemoradiotherapy has been recommended as the standard treatment 1 . Even after intensive treatments, patients remain at high risk of recurrence or metastasis 2 . In recent years, neoadjuvant therapy before surgery has been shown to reduce the burden of locoregional disease, resulting in improved surgical outcomes; to reduce the risk of distant metastases; and to predict prognosis based on the pathological response in various solid tumors 3 . However, its role remains ambiguous in OSCC. Neoadjuvant chemotherapy using cisplatin plus uorouracil (PF) or docetaxel plus cisplatin plus uorouracil (TPF) regimen has been explored in patients with OSCC, but has not demonstrated survival bene ts beyond standard treatment 4,5 . Thus, exploring effective neoadjuvant therapeutic approaches for locally advanced resectable OSCC remains an urgent need.
Immune checkpoint blockade has been demonstrated to have clinically meaningful anti-tumor activity in recurrent/metastatic head and neck squamous cell carcinoma (HNSCC, including OSCC) 6, 7 . Preclinical data suggest that while the tumor is in place, neoadjuvant immunotherapy hypothetically leverages the higher levels of endogenous tumor antigens, thereby enhancing T cell priming and resulting in stronger effects than those of adjuvant therapy 8 . In the neoadjuvant setting, immune checkpoint blockade has shown promising results against many other tumor types [9][10][11][12] . Targeted drugs against vascular endothelial growth factor receptor (VEGFR) or angiogenesis have shown to relieve immunosuppression through blood vessel normalization and the oxygen metabolism pathway, thereby having a synergistic effect with anti-programmed cell death-1 (PD-1) immunotherapy and concurrently diminishing the risk of immune-related adverse effects [13][14][15][16] . The combination of camrelizumab (an anti-PD-1 antibody) and apatinib (a VEGFR inhibitor) has shown favorable anti-tumor activity and manageable safety in various types of advanced cancers [17][18][19][20] .
In this study, we conducted a prospective, single-arm trial to assess the safety and pathological e cacy of a chemo-free combination of camrelizumab and apatinib as neoadjuvant therapy in patients with locally advanced resectable OSCC.

Patient information
From April to December 2020, 21 patients were enrolled, and one patient withdrew at the beginning of treatment. The characteristics of the 21 enrolled patients are listed in Table 1. Twenty patients received radical surgery, and 18 patients received adjuvant radiotherapy (Fig. 1).

Safety
The most common neoadjuvant therapy-related AEs were hyperbilirubinemia (N = 8, 40%), thrombocytopenia (N = 7, 35%) and proteinuria (N = 6, 30%). No neoadjuvant therapy-related grade 3 or above AEs were detected ( Table 2, Supplementary Table S1 and S2 in Additional le 4). One patient postponed the second cycle of camrelizumab for 14 days because of grade 2 thrombocytopenia, and one patient suspended apatinib for 21 days because of grade 2 hyperbilirubinemia. Surgery-related AEs occurred in four patients, including one each of subcutaneous exudate, post-tracheostomy bleeding, postap-reconstruction pharyngeal stula, and wound infection. The post-tracheostomy bleeding was due to unsecured ligation of the anterior jugular vein, which was detected during the surgical exploration. The other three patients showed no AEs during preoperative laboratory tests, and their surgery-related AEs were all controlled within 2 weeks and were deemed to be unrelated to the neoadjuvant therapy. Two severe AEs occurred: one patient experienced unexplainable cardiac troponin I elevation, resulting in a surgery delay for 7 days, then recovering within 1 week without any corticosteroid treatment; the other patient experienced unexplained shock after radiotherapy and died.

E cacy
The pathological e cacy indicated that eight (40%) patients achieved MPR including one patient (5%) who achieved pathological complete response (pCR). Among them, all ve patients with CPS Radiographic responses according to RECIST 1.1 were performed on basis of imaging examinations before and after neoadjuvant therapy. The radiographic response indicated three patients with partial response (PR), ten patients with stable disease (SD), and six patients with progressive disease (PD) (Supplementary Table S4 in Additional le 4). One super cial gingival lesion was undetectable on radiographic examinations and thus was not evaluated. Interestingly, among the eight patients who achieved MPR, only three showed radiographic PR ( Fig. 2A). One radiographic PD lesion was further pathologically con rmed MPR. All patients with PD lesions received surgery, and no recurrence was found in primary sites. For the regional metastatic lymph nodes, pathological response was found in 60% (6/10) of patients, with the characteristics of necrosis, multinucleated giant cells, and calci cation. In the patient who achieved pCR in the primary tumor, pCR in one lymph node was also observed (Supplementary Table S5 in Additional le 4).

Pathological response characteristics
We systematically reviewed the pathological features in resected specimens and proposed immunerelated pathological response criteria (irPRC) for neoadjuvant therapy in OSCC. We observed the following characteristics of the immune-related pathological regression bed in OSCC: multinucleated giant cell in ltration, dystrophic calci cation, tumor-in ltrating lymphocytes, foamy macrophages, neovascularization, proliferative brosis, tertiary lymphoid structure, and dense plasma cells (Additional le 2). In two patients who achieved MPR, the tertiary lymphoid structure was found in the tumors after neoadjuvant therapy (Additional le 2).
No signi cant change in the degree of TIL in ltration was found over the neoadjuvant therapy course (Supplementary Fig. S1 in Additional le 4). The characteristics of TIL in ltration in surgically resected tumors between two groups were further compared, and the MPR group showed more CD4+ T cell in ltration and a higher CD8+/FoxP3+ ratio than the non-MPR group (Fig. 3). One patient from the non-MPR group was found to have tumor progression on radiographic evaluation, with a change in growth kinetics exceeding 50% and new neck lymph node metastasis, and was con rmed to have disease hyperprogression (HPD, Fig. 4A, B). Interestingly, for this patient, high levels of in ltration for CD8+ and CD163+ cells were found in the biopsy, and CD8+ was diminished while CD163+ was signi cantly elevated in the surgical sample ( For angiogenesis evaluation, decreased CD31 (a marker of vascular endothelial cells) and -SMA (a marker of pericytes) expression were found after neoadjuvant therapy, thus con rming the antiangiogenesis effect in tumors ( Fig. 5A and B). No signi cant difference in CD31 or -SMA expression was found between the MPR and non-MPR groups (Fig. 5C).

Discussion
Our results provided the rst evidence that neoadjuvant therapy using a chemo-free combination of camrelizumab and apatinib was well tolerated in patients with OSCC, with an MPR of 40%.
The safety pro le of camrelizumab and apatinib in the neoadjuvant setting was mostly consistent with the previously reported in advanced cancers 17,18,21 , We observed no grade 3-4 neoadjuvant therapyrelated AEs, which might be due to a short course of administration of camrelizumab and apatinib.
Furthermore, the safety in our trial seems to be superior to that of other neoadjuvant chemotherapy regimens in OSCC trials, such as neoadjuvant chemotherapy with TPF regimen (9-38% grade 3-4 therapy-related AEs) 4 In all these reported neoadjuvant immunotherapy trials in OSCC or HNSCC, the pathological response varied in terms of MPR ratios, and the assessment procedures also varied. Unlike non-small-cell lung carcinoma and melanoma 35,36 , in OSCC or HNSCC, irPRC has not been well de ned. In previous neoadjuvant immunotherapy trials, the pathological response has been described as "visible regressed tumor, in ammation, giant cell reaction and acellular keratin" and quanti ed as "a percentage of the overall tumor bed (area of pathological response/area of pathological response plus viable tumor)" 27,28 .
Regardless of the assessment method being used, the de nition of the tumor regression bed after neoadjuvant therapy is key, especially in signi cantly shrunk tumors. On basis of the criteria of determining the range of the immunotherapy-induced tumor regression bed proposed in lung cancer 36 . We systematically evaluated the features in tumors from different oral cavity sites in this trial, thus providing a reference of irPRC for following neoadjuvant immunotherapy in OSCC.
The irPRC we proposed for OSCC did not include the pathological response characteristics for neck lymph node metastasis. We found only one lymph node with con rmed tumor regression, whereas a different reaction occurred in another lymph node from the same patient. As described in a previous lung cancer study, the limitations in nodal disease assessment have been attributed to sampling issues. 36 In agreement with ndings from the study using nivolumab, a similar response has been found between primary sites and lymph nodes 37 . Nodal upstaging occurred in four patients in our study, and all the corresponding primary tumors did not achieve MPR; therefore, lymph nodes are suggested to be monitored more frequently during neoadjuvant treatment.
Because of the short period of 2 weeks between the last neoadjuvant immunotherapy and surgery, radiographic re-evaluation based on modi ed RECIST1.1 criteria for immune based therapeutics (iRECIST) could not be performed in this study 38 . In pathological re-evaluation of the resected lesions, the RECIST 1.1 criteria did not show su cient sensitivity in response assessment in our trial. Among the eight patients who achieved MPR, only three showed PR on radiographic scans. One radiographic PD lesion in our trial was further pathologically con rmed MPR, thus indicating the importance of reevaluation on progression assessment after immunotherapy. This nding was consistent with the radiographic response analysis of neoadjuvant immunotherapy in lung cancer, in which 30% of patients with SD virtually achieved pCR 39 . In future neoadjuvant therapy trials, other modi ed methods for radiographic response evaluation should be proposed, such as the criteria used in the window of opportunity, in which a size reduction 10%, rather than 30%, might be de ned as indicating a "radiographic responder" 37 .
In agreement with ndings from other neoadjuvant immunotherapy trials in OSCC or HNSCC 27, 28 , the degree and the features of baseline TIL in ltration did not predict pathological response in our study. However, we observed more CD4+ T cell in ltration and a higher CD8+/FoxP3+ ratio in resected tumors that achieved MPR. In addition, an abnormal increase of CD163+ in ltration was observed in HPD tumors. Because higher levels of CD4+ and CD8+/FoxP3+ ratios were associated with better outcomes of neoadjuvant therapy, and M2 macrophages in the tumor microenvironment have been reported to be associated with the occurrence of HPD [40][41][42] , further analyses of our tumor samples are urgently needed to explore the underlying mechanisms. Angiogenesis markers for apatinib treatment, including CD31 and -SMA, were inhibited by neoadjuvant therapy in this trial, similarly to previously reported preclinical results 43,44 . The expression of angiogenesis markers showed no differences between the MPR and non-MPR groups, whereas patients with high CPS tended to have a higher rate of MPR. Therefore, in this trial, we speculated that the pathologic response might be mainly due to the anti-PD-1 effect rather than antiangiogenesis. However, crosstalk between tumor vascular normalization and immune reprogramming exists and plays a vital role in pathologic response 45 . A deeper understanding of this synergic effect could provide novel strategies to predict responses and to further improve the e cacy of cancer immunotherapy.
In the past decade, MPR and pCR have been considered a candidate early surrogate endpoint for survival in the neoadjuvant setting 35,46,47 ; however, whether MPR or even pCR might result in long-term survival improvement remains to be con rmed in this and other neoadjuvant immunotherapy trials. For patients with OSCC who achieved MPR or even pCR through neoadjuvant therapy, whether the intensity of surgery and the following adjuvant therapy could be reduced remains controversial. The pathological response characteristics in lymph nodes, as well as their potential correlation with the e cacy of immunotherapy, require further investigations. To investigate the predictive biomarkers for re ned neoadjuvant therapy, further in-depth biomarkers in the host, tumor, and blood (such as PET-CT parameters, tumor mutation burden, neoantigens, circulating tumor DNA, and T cell activation and exhaustion) should be evaluated in this and other ongoing neoadjuvant trials 48 .
In conclusion, this pilot trial showed that neoadjuvant therapy using a chemo-free combination of camrelizumab and apatinib was well tolerated in patients with OSCC. The MPR rate was promising and CPS might be a signal predictor. These results support further neoadjuvant therapy trials for OSCC using anti-PD-1 plus anti-VEGFR, guided by the CPS.

Procedures
The patients received three cycles of intravenous camrelizumab (200 mg) on d1, d15, and d29; and oral apatinib (250 mg) daily, starting on d1 and ending on the 5th day before surgery. Standard radical surgery was planned on d42-45. Adjuvant radiotherapy or chemoradiotherapy was planned within 6 weeks after surgery, according to the pathological stage.
The standard operation procedure of determining the regression bed induced by neoadjuvant immunotherapy in oral cancer was proposed (Additional le 2). The percentage of residual viable tumor (RVT) cells was evaluated on resected tumor slides. Radiographic response evaluation was performed according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 criteria.

Study endpoints
The primary endpoints were safety and major pathological response (MPR) rate (MPR, de ned as the presence of 10% or fewer RVT cells).

Statistical analyses
A sample size of 20 evaluable patients was required to achieve 90% power to detect an increase in the MPR rate from 7% (anti-PD-1 monotherapy) to 30%, with a one-sided exact test with a signi cance level (alpha) of 0.0500.
Overall survival from the date of enrollment until death was evaluated with the Kaplan-Meier method. Based on the different CPS cutoffs, the P value of patient number between MPR and non-MPR groups was calculated with the chi-square test. The con dence interval (CI) of the survival rate and local recurrence rate were calculated with the Clopper-Pearson method, and the P values of quanti ed uorescence differences between two groups was calculated with Student's t-test. The signi cance level for two-sided P values was set at 0.05 in statistical analyses. Statistical analysis was performed in IBM SPSS Statistics and GraphPad Prism software.

Ethics approval
The trial followed the ethical guidelines of the Declaration of Helsinki and was approved by the Institutional Ethics Committee, Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. Each patient provided signed informed consent before participating in this trial.

Data availability
The trial protocol is available as Additional le 1 in the Supplementary Information le. Any additional datasets used and/or analysed during the current trial are available from the corresponding author on reasonable request.      Comparison of TILs between MPR and non-MPR groups. In ltration density of different lymphocytes subgroups, CD8+/CD163+ ratio, and CD8+/FoxP3 ratio in tumor samples before-and after-neoadjuvant therapy (blue dots for the MPR group, red dots for the non-MPR group). Statistical signi cance was denoted by *p < 0.05.

Figure 4
Features of the hyperprogression disease. In the NO.14 patient who showed hyperprogression disease: