Neoadjuvant chemo-immunotherapy with camrelizumab plus nab-paclitaxel and cisplatin in resectable locally advanced squamous cell carcinoma of the head and neck: a pilot phase II trial

Abstract

Neoadjuvant chemoimmunotherapy has emerged as a potential treatment option for resectable head and neck squamous cell carcinoma (HNSCC). In this single-arm phase II trial (NCT04826679), patients with resectable locally advanced HNSCC (T2‒T4, N0‒N3b, M0) received neoadjuvant chemoimmunotherapy with camrelizumab (200 mg), nab-paclitaxel (260 mg/m²), and cisplatin (60 mg/m²) intravenously on day one of each three-week cycle for three cycles. The primary endpoint was the objective response rate (ORR). Secondary endpoints included pathologic complete response (pCR), major pathologic response (MPR), two-year progression-free survival rate, two-year overall survival rate, and toxicities. Here, we report the perioperative outcomes; survival outcomes were not mature at the time of data analysis. Between April 19, 2021 and March 17, 2022, 48 patients were enrolled and received neoadjuvant therapy, 27 of whom proceeded to surgical resection and remaining 21 received non-surgical therapy. The ORR was 89.6% (95% CI: 80.9, 98.2) among 48 patients who completed neoadjuvant therapy. Of the 27 patients who underwent surgery, 17 (63.0%, 95% CI: 44.7, 81.2) achieved a MPR or pCR, with a pCR rate of 55.6% (95% CI: 36.8, 74.3). Treatment-related adverse events of grade 3 or 4 occurred in two patients. This study meets the primary endpoint showing potential efficacy of neoadjuvant camrelizumab plus nab-paclitaxel and cisplatin, with an acceptable safety profile, in patients with resectable locally advanced HNSCC.

Fig. 1. Study flow chart.

In total, 48 patients were enrolled in this trial and received neoadjuvant therapy; among them, 21 patients received adjuvant therapy without surgery, 27 patients underwent surgical resection.

Fig. 2. The waterfall plot of best radiographic response by RECIST 1.1 (n = 48).

Each bar indicates one patient. Source data are provided as a Source Data file.

Fig. 3. Correlation analysis of radiographic response and pathologic response (n = 26).

Sankey plot shows the relationship between radiographic response and pathologic response (a). The consistency between radiographic response and pathologic response analyzed by using the two-sided Fisher exact test (b). Source data are provided as a Source Data file.

Fig. 4. The genetic landscape and radiographic tumor response analysis.

Alterations as assessed by next-generation sequencing of baseline primary tumor samples (n = 47) (a). Comparisons of radiographic tumor response between HPV-positive (n = 9) and HPV-negative (n = 39) patients (P = 0.012) (b); TP53-mutant (n = 37) and TP53-wild-type (n = 10) patients (P = 0.006) (c); and TERT-mutant (n = 11) and TERT-wild-type (n = 36) patients by using the two-sided Mann–Whitney U nonparametric test (P = 0.01) (d). Correlation between HPV status (HPV-positive: n = 9; HPV-negative: n = 39) and TP53 (n = 37, P < 0.0001) or TERT mutations (n = 11, P = 0.16) by using the two-sided Mann–Whitney U nonparametric test (e). Comparisons of radiographic tumor response between TP53-mutant (n = 35) and TP53-wild-type (n = 3) patients (P = 0.807) (f); and TERT-mutant (n = 11) and TERT-wild-type (n = 27) patients (P = 0.049) (g) in HPV-negative patients by using the two-sided Mann–Whitney U nonparametric test. The histogram plots show the mean values of radiographic tumor response and standard deviation (SD). The dot represents an individual data point. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns no significance. Source data are provided as a Source Data file.

Fig. 5. Tumor-infiltrating lymphocyte analysis.

The radiographic tumor response was significantly associated with the densities of tumor-infiltrating CD8 + T cells (n = 47, Spearman R = 0.405, P = 0.0005) (a) and tumor-infiltrating M1-like macrophage cells (n = 47, Spearman R = 0.375, P = 0.009) (b) by using the two-sided Spearman correlation text. The line represents a fitted line. The gray shadow represents the corresponding 95% confidence interval. The baby-blue dot represents an individual data point. The multiplex immunofluorescence images of the tumor sites in patient 20 (c) and patient 45 (d). Primary antibodies targeting CD8, CD56, CD68, HLA-DR, and Pan-CK were used on the same slide. Comparisons of the densities of tumor-infiltrating CD8+ cells (P < 0.0001) and M1-like macrophage cells (P = 0.01) between HPV-positive (n = 9) and HPV-negative (n = 38) patients (e); CD8+ cells (P = 0.0002) and M1-like macrophage cells (P = 0.0028) between TP53-mutant (n = 37) and TP53-wild-type (n = 10) patients (f), and CD8+ cells (P = 0.19) and M1-like macrophage cells (P = 0.07) between TERT-mutant (n = 11) and TERT-wild-type (n = 36) patients (g) analyzed by using the two-sided unpaired t test. Box-and-whisker plots show the distribution (box, whiskers, and outliers). The center line represents the median, the box limit represents the interquartile range (IQR), the whiskers represent the 1.5×IQR, and the outliers represent an individual data point. *P < 0.05, **P < 0.01, ***P < 0.001, ns no significance. Source data are provided as a Source Data file.

Fig. 6. Exploratory analysis of pathologic response characteristics.

The percentages of patients with HPV infection (a, P = 0.03), PD-L1 positive expression (b, P = 0.046), or TP53 mutation (c, P = 0.18) in the pCR (n = 15) and non-pCR (n = 11) patients were compared by using the two-sided Fisher exact test. Comparison of the density of tumor-infiltering immune cells in patients with pCR (n = 15) and non-pCR (n = 11) by using the two-sided Wilcoxon rank-sum test (d). Box-and-whisker plots show the distribution of cell density. The center line represents the median, the box limit represents the interquartile range (IQR), the whiskers represent the 1.5×IQR, and the outliers represent individual data point. *P < 0.05, ns no significance. Source data are provided as a Source Data file.