Pretreatment circulating tumor HPV DNA level is associated with nodal disease burden, HPV genotype, and lymphovascular invasion among patients with HPV-associated oropharynx carcinoma

Presentation: A116
Topic: Oropharynx / HPV Related Disease
Type: Poster
Authors: Annette A Wang, MS1; Glenn J Hanna, MD2; Evan Carey2; Eleni M Rettig, MD3
Institution(s): 1Harvard Medical School; 2Dana-Farber Cancer Institute; 3Brigham and Women's Hospital Otolaryngology-Head and Neck Surgery


Circulating tumor human papillomavirus (HPV) DNA (ctHPVDNA) is a dynamic biomarker for HPV-associated oropharynx squamous cell carcinoma (HPV+OPSCC) that varies with burden of disease and treatment response. Determinants of pre-treatment ctHPVDNA level are not well described. This single institution cross-sectional study examined clinicopathologic characteristics associated with pre-treatment ctHPVDNA levels among individuals with HPV+OPSCC.

Methods: Eligible cases were incident or recurrent HPV+OPSCCs diagnosed December 2019-August 2021 evaluated at a tertiary academic center with pre-treatment ctHPVDNA testing. Tumors were considered HPV+OPSCCs if they were p16-positive (p16+) by immunohistochemistry and/or high-risk HPV-positive (HR-HPV+) by in-situ hybridization.

Digital droplet PCR analysis of tumor tissue-modified HPV DNA (NavDx®) from HPV genotypes 16/18/31/33/35 was performed by Naveris (Natick, MA), an independent CLIA-certified laboratory, and reported as fragments/ml (frag/ml) of plasma ctHPVDNA. Cases with ‘indeterminate’ ctHPVDNA, reported as 5-12 frag/ml, were considered detectable and assigned values of 8.5 frag/ml. Clinicopathologic characteristics were ascertained via electronic medical record abstraction. AJCC 8th edition clinical staging was recorded and adjusted to reflect disease extent at time of blood collection in cases of excisional diagnostic procedures. Proportions were compared using Fisher’s exact tests, and medians using Wilcoxon rank-sum or Kruskall-Wallis tests.

Results: The study comprised 84 patients. Most were men (N=74, 87%), white (N=79, 94%), with median age=62.7 years. ctHPVDNA was positive in 72 patients (86%), indeterminant in 2 (2%) and undetectable in 10 (12%). Among undetectables, 8 tumors were p16+/HR-HPV+, and 2 were p16+ without available HR-HPV testing. ctHPVDNA genotype was most commonly HPV16 (N=64, 86%), followed by HPV33 (N=7, 9%), HPV35 (N=2, 3%), and HPV31 (N=1, 1%). Nodal metastasis was significantly associated with detectable ctHPVDNA. Most individuals with N0 disease had undetectable ctHPVDNA (6 of 9, 67%), compared with only 4 of 75 (5%) individuals with N1-3 disease (p=0.001).

Median ctHPVDNA level overall was 284 frag/ml (range=0-60,061; interquartile range [IQR]=27-2,547). ctHPVDNA level was higher for men (289 frag/ml, IQR=33-3,690) than women (68 frag/ml, IQR=9-110; p=0.016), and increased with higher N stage (N0: 0 frag/ml, IQR=0-9; N1: 344 frag/ml, IQR=68-2,407; N2: 1,297 frag/ml, IQR=140-3,566; N3: 3,339 frag/ml, IQR=22-6,656; p=0.001 and ptrend=0.001). Non-HPV16 genotype of ctHPVDNA was associated with significantly lower levels of ctHPVDNA (median=112 frag/ml, IQR=23-237) when compared with HPV16 (median=850 frag/ml, IQR=101-4,255; p=0.012). Demographic characteristics, smoking status, tumor site, T stage, clinical extranodal extension, and neutrophil:lymphocyte ratio were not associated with ctHPVDNA level. Among patients treated surgically, although cases were limited (N=19), ctHPVDNA was significantly higher among those with (N=10) versus without (N=9) lymphovascular invasion (LVI; median=8.5 frag/ml, IQR=0-9 versus 274 frag/ml, IQR=142-10,292; p=0.014), but was not associated with perineural invasion or pathologic extranodal extension.

Conclusion: ctHPVDNA is detectable among nearly 90% of p16+ OPSCC patients, with widely variable levels that appear higher among those with greater nodal disease burden, consistent with prior reports. We also observed higher ctHPVDNA among patients with HPV16 versus non-HPV16 ctHPVDNA, supporting emerging evidence for HPV+OPSCC heterogeneity by HPV genotype. Finally, we report an association between higher ctHPVDNA and LVI, implicating LVI as a means for circulatory shedding of tumor DNA.