Background

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer in the world, with 650,000 new cases diagnosed each year. Tobacco use, heavy alcohol intake, and carcinogenic human papillomavirus (HPV) are strong risk factors [1, 2]. HPV-positive HNSCC responds favorably to chemoradiation leading to improved prognosis [3, 4].

Patients who have locoregionally advanced disease at diagnosis receive combined modality therapy with a platinum-based chemotherapy backbone; [5, 6] however, in 2006, cetuximab, a chimeric monoclonal antibody and an epidermal growth factor receptor (EGFR) inhibitor, was approved with radiation therapy for locoregionally advanced HNSCC. The addition of cetuximab to radiotherapy improved survival in HNSCC [7]. Cetuximab has been proposed as an alternative to platinum chemotherapy due to a more favorable toxicity profile and improved tolerability in those with renal disease or in the elderly. However, two recent multi-centers randomized clinical trials in patients with HPV associated HNSCC have demonstrated both inferior overall survival of patients who received cetuximab based instead of cisplatin-based radiotherapy and no difference in terms of toxicity profile [8, 9]. One hypothesis for the decreased efficacy of cetuximab in HPV associated HNSCC is the decreased dependence on the EGFR pathway compared to HPV-negative tumors [10].

We aimed to compare real-world outcomes of cisplatin-based versus cetuximab-based primary chemoradiation in an academic setting at Loma Linda University Medical Center. We also re-analyzed the HNSC dataset from The Cancer Genome Atlas (TCGA) [11] to clarify the relationship between HPV and EGFR statuses in HNSCC.

Materials and methods

Data collection

ICD-9, ICD-10, and CPT codes were used to search the Loma Linda University Health Cancer Registry to find patients diagnosed with HNSCC between 2006 and 2018. This study was approved by Loma Linda University Medical Center Institutional Review Board (IRB Approval #: 5180298). The following data were obtained for each patient: age, sex, smoking status, primary malignancy site, pathology report, staging, type of radiation and number of radiation treatments, chemotherapy regimen, and number of chemotherapy cycles completed. Radiation dose ranged between 48 and 72 Gy.

Patient selection

Eligible patients were between the ages of 18–99 years who had locally advanced head and neck cancer with initial stages III-IVB receiving concurrent chemoradiation with either cisplatin or cetuximab as first-line therapy. Primary tumors of the oral cavity, oropharynx, and hypopharynx/larynx were included. We excluded those who underwent surgery as first-line therapy, concurrent chemoradiation as second-line or salvage therapy for recurrent or metastatic disease, underlying malignancy other than head and neck cancer, those who did not receive either cisplatin or cetuximab as part of chemotherapy regimen, and nasopharyngeal tumors.

Cases that met inclusion criteria were then separated by chemotherapy received. Categorical variables analyzed included smoking and HPV/p16 statuses, which was abstracted from pathology reports consistent with College of American Pathologists recommendation [12]. p16INK4a immunohistochemistry stain and HPV in situ hybridization were performed in Clinical Laboratory Improvement Amendments regulated laboratories with in vitro diagnostic antibodies. Patients with a smoking history were further categorized into light smoking (< 10 pack-years) and heavy smoking (≥ 10 pack-years). Comorbidities was considered positive if patients had ≥ 2 comorbidities and negative if < 2.

Endpoints and statistics

Primary endpoint was overall survival (OS) by treatment with either cisplatin or cetuximab. OS was analyzed by the Kaplan–Meier and log-rank methods. Survival was compared by univariate Cox proportional hazards model between treatment groups. Survival analyses were stratified by HPV/p16 and smoking statuses. Statistical analysis with p-values of < 0.05 was considered significant. Statistics were performed with R packages “survival” and “survminer” (Vienna, Austria).

TCGA data and analysis

Publicly available data for clinical and phenotypic information, gene expression (by RNA sequencing), and copy number (generated by the Affymetrix Genome-Wide Human SNP Array 6.0 platform) data of the HNSC cohort from TCGA were obtained from University of California Santa Cruz Xena Browser. Gene expression data were represented as RSEM scores, which were log2-transformed normalized counts [13]. The copy number estimated values are thresholded to −2, −1, 0, 1, 2, representing homozygous deletion, single copy deletion, diploid normal copy, low-level copy number amplification, or high-level copy number amplification reported by GISTIC2 [14, 15]. EGFR copy number statuses were grouped to amplified (GISTIC2 score ≥ 1) and non-amplified (GISTIC2 score ≤ 0). HPV status was obtained from the supplementary information of the original Nature article [11], and most of them were ascertained by the presence of HPV transcripts or genomes by whole genome/exome and RNA sequencing. All analyses were performed on KNIME Analytics.

Results

Patient and tumor characteristics

A total of 1545 patients were retrieved from the LLUMC Cancer Registry between 2006 and 2018. Of these patients, 121 met inclusion criteria and were further divided into cisplatin (n = 66) and cetuximab (n = 55) cohorts (Fig. 1). A summary of demographic information can be found in Table 1.

Fig. 1
figure 1

Patient screening flowchart

Full size image
Table 1 Patient characteristics based on primary chemotherapy received
Full size table

Cisplatin chemoradiation improves OS

The median OS for patients receiving cisplatin was not reached compared to 132 months in the cetuximab group (p = 0.03, Hazard Ratio [HR] = 0.44, [95% confidence interval (CI), 0.20–0.95]) (Fig. 2a).

Fig. 2
figure 2figure 2

Overall survival of (a) the overall study population, (b) the HPV/p16-positive population, (c) the smoking population, and (d) the advanced age population that received primary chemoradiation with cisplatin versus cetuximab

Full size image

Improved OS in HPV/p16-positive cancer treated with cisplatin-based chemoradiation

Among the patients who were HPV/p16-positive (n = 55), those who received cisplatin chemoradiation had longer OS. The median OS for patients with HPV/p16-positive cancer treated with cisplatin-based therapy was not reached compared to 60 months for HPV/p16-positive cancer treated with cetuximab (p = 0.036, HR = 0.25 [95% CI 0.061–1.00]) (Fig. 2b).

Cisplatin chemoradiation improved OS in patients with positive smoking history

In patients with a smoking history (n = 76), cisplatin-based radiotherapy was associated with longer OS. The median OS in the cisplatin group was not reached, while that in the cetuximab cohort was 60 months (p = 0.03, HR = 0.36 [95% CI 0.14−0.93]) (Fig. 2c).

Age and outcomes

Within our study population, patients in the cisplatin group were significantly younger than those in the cetuximab group (56.80 versus 60.04, respectively, p = 0.0003). Of patients over 60 years old (n = 51), those treated with cisplatin had improved OS (p = 0.035, HR = 0.23, [95% CI 0.051−1.03]) (Fig. 2d).

TCGA HNSC data re-analysis

Within TCGA who had HPV-positive cancers, none of the oropharyngeal (n = 22) and only one of the overall head and neck patients (n = 36) had the amplified EGFR gene region, whereas within the HPV-negative cohort, six oropharyngeal (n = 11, 54.5%) and 124 of the overall head and neck patients (n = 243, 51.0%) had the amplified EGFR gene region. Fisher’s exact tests indicated that the HPV-negative cohort is more likely to possess EGFR amplification in head and neck patients (p < 0.0001). At the transcription level, RNA expressions of the EGFR gene in HPV-negative cancers were 1.619 times (95% CI 1.22 to 2.14) higher than HPV-positive cancers in the overall head and neck cohort (p < 0.0001) (Fig. 3).

Fig. 3
figure 3

Comparisons of EGFR expression levels of HNSCC cases in TCGA HNSC cohort based on HPV status and amplification (Amp) status of the EGFR gene region (x-axis). The EGFR expression level (y-axis) by RNA-sequencing is presented in RSEM scores [13] on a log2 scale

Full size image

Discussion

In this single institution retrospective study, we investigated real-world outcomes of locoregionally advanced HNSCC treated with primary chemoradiation with cisplatin versus cetuximab.

Our finding of an OS benefit in HPV/p16-positive cases treated with cisplatin-based therapy compared to cetuximab is consistent with recent phase 3 clinical trial data [8, 9] and national database studies [16,17,18]. While our study and the VA study by Bauml et al. both found that cetuximab with radiation yields inferior OS, more than 99% of the VA study population are male and it did not perform a subgroup analysis of HPV status. This growing evidence supports cisplatin-based radiotherapy as the standard of care for patients with HPV/p16-positive cancer with low risk (e.g. non-smokers or < 10 pack-years smoking history) [9]. We also found the survival benefit with cisplatin therapy to be consistent among those who smoke and those with advanced age. However, cetuximab likely still will play a role in patients with poor renal function, as both even though head to head trials suggested no different overall toxicity, cisplatin consistently demonstrated increased rates of acute kidney injury [8, 9].

Some authors have proposed that the inferior survival outcomes in patients with HPV associated oropharyngeal HNSCC who received concurrent chemoradiation with cetuximab instead of cisplatin might be related to differences in EGFR expression and HPV status. Indeed, we observed that amplification of the EGFR gene region was rare to absent in HPV-positive cancers, while about half of the HPV-negative cases possessed different degrees of amplifications in the TCGA HNSC dataset. Expression of the EGFR gene was also significantly lower in HPV-positive HNSCC, consistent with prior reports [19]. However, EGFR expression has not been shown to be predictive of clinical benefit for anti EGFR therapy in SCCHN, which may be related to the lack of correlation between EGFR expression and autophosphorylation activity [20, 21].

The main limitation of this study is that, as a retrospective study, we cannot exclude possible confounding factors that may contribute to inferior OS of cetuximab-based radiotherapy in patients with HPV/p16-positive cancers, such as insufficient renal function that have precluded a cisplatin regimen or the selection bias of cetuximab in elderly and frail patients. However, based on our data even the population with advanced age still benefited from high dose cisplatin, and further more patients with at least 2 comorbidities received cisplatin treatment. In conclusion, our result was consistent with the recent prospective randomized clinical trials, which have provided strong evidence of the superiority of cisplatin compared with cetuximab in HPV/p16-positive cancers [8, 9].