Accuracy of Imaging Modalities for Detecting Extracapsular Spread of Cervical Lymph Node Metastases in HPV-Associated Oropharyngeal Cancer: A Systematic Review

Abstract

Background: Extracapsular spread (ECS) of lymph node metastases is associated with poor prognosis and its detection in head and neck cancer (H&NC) is crucial for treatment planning. Commonly used imaging modalities to detect ECS in H&NC include computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET) and ultrasonography (US). Currently there is no gold standard imaging modality to detect ECS in H&NC, leaving clinicians to rely heavily on clinical examination and surgical histopathology for ECS based treatment decisions. The purpose of this study was to undertake a systematic review using Joanna Briggs Institute (JBI) methodology that aimed at identifying and synthesising the best available evidence regarding the accuracy of conventional imaging modalities and their abilities to detect ECS in the specific population group of patients with human papillomavirus positive (HPV+) oropharyngeal squamous cell carcinoma (OPSCC). Inclusion criteria: Type of participants: Participants were those with a confirmed diagnosis of HPV+ OPSCC and suspected diagnosis of cervical lymph node metastases and ECS. Participants were not excluded due to age, sex, race or education status. Type of index tests: This review examined studies that utilised a conventional imaging modality to detect radiologic ECS in HPV+ OPSCC. Type of reference test: This review examined studies that utilised surgical histopathology as the reference standard for the diagnosis of ECS (gold standard for ECS detection). Type of outcomes: This review examined two primary and four secondary outcomes of interest. The primary outcomes of interest included: sensitivity and specificity measures with 95% confidence intervals for each imaging modality used to detect ECS in HPV+ OPSCC. The secondary outcomes included: positive predictive value (PPV), negative predictive value (NPV), area under the ROC curve (AUC) and interobserver agreements (K) (where applicable) for the different imaging modalities. Diagnosis of interest: The phenomena of interest in this review was ECS of cervical lymph node metastases (also known as extra-nodal extension (ENE)). Type of studies: This review examined published studies that examined the diagnostic accuracy (including sensitivity and specificity) of an imaging modality used to detect ECS in HPV+ OPSCC. Diagnostic cohort studies were the preferred study design for inclusion. All six of the included studies were retrospective cohort studies. Methods: Methodological approach: The methodological approach to the review was based on JBI guidance for systematic reviews involving diagnostic test accuracy (DTA) studies. Search strategy: A comprehensive search using a three-phased approach was conducted across four databases, one clinical trials register, as well as a manual search for primary studies (published) in the reference lists of all included studies. There was no restriction on publication date, however. only studies in English were included in the review. Methodological quality: Two reviewers assessed the methodological quality of the included studies using the QUADAS-2 tool. The QUADAS-2 tool is structured around assessing for risk of bias in four domains; Patient Selection, Index Test, Reference Standard, and Flow and Timing. Data extraction: Quantitative data was extracted using the JBI data extraction tool for DTA studies. Data analysis Meta-analysis and assessment of heterogeneity was conducted on four CT studies using a random-effects model. The remaining two studies underwent a narrative synthesis. The GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) approach was used to assess the certainty in the evidence. Results: Out of 1772 hits, six retrospective cohort studies were included in the review and four underwent meta-analysis. Four investigated the diagnostic ability of CT, one investigated PET/CT, and one investigated 'CT and MRI' (with no separation of index test to outcomes). Meta-analysis of the four CT studies showed CT had an overall sensitivity of 77% (60-94%) and specificity of 60% (47-73%). PET/CT had a sensitivity of 86% (73-94%) and specificity of 76% (61-87%). 'CT and MRI' had a sensitivity of 62% (53-70%) and specificity of 78% (70-84%). No meta-analysis or comparison meta-regression could be performed on the PET/CT or 'CT and MRI' studies. Conclusions: The findings of this review imply pooled CT specificity values (60%) are too low to suggest clinical value for CT as a diagnostic tool to detect ECS in HPV+ OPSCC. Pending further research, the use of CT and PET/CT however might have clinically acceptable sensitivity and negative predictive values to help confirm the absence of radiologic ECS in HPV+ OPSCC. No studies on MRI or US were identified for assessment of ECS in HPV+ OPSCC. Implications for practice: There is insufficient evidence to suggest CT or PET/CT are reliable diagnostic tools for radiological detection of ECS in HPV+ OPSCC. Further studies of high-quality involving CT, PET/CT, MRI and US are required to help establish clinical guidance and a gold standard for radiologic ECS detection in HPV+ OPSCC.