MRI Posttreatment Surveillance for Head and Neck Squamous Cell Carcinoma: Proposed MR NI-RADS Criteria

Study Design and Patient Selection

Our institution's ethics committee approved this single-center, retrospective study and waived the requirement for informed consent. Data were retrieved from the medical records and PACS. This study included 69 patients with head and neck squamous cell carcinoma (HNSCC) who were referred for posttreatment MR imaging surveillance between June 2018 and September 2020. Patients with histopathologically diagnosed primary HNSCC who underwent CEMRI surveillance within the first year after treatment, starting from 8–10 weeks after completing their treatment course, were included in this study. We considered the first posttreatment MR imaging scan, irrespective of whether it was the baseline scan or preceded by CECT with or without PET. We excluded patients with inadequate clinical or imaging follow-up or who had missing data, NI-RADS 4 with confirmed recurrence before scanning, noncontrast-enhanced scans, and scans with low-quality images that showed remarkable artifacts.

An initial search in the PACS revealed that 335 CEMRI neck scans were performed for 221 patients between June 2018 and September 2020, of whom 87 patients had HNSCC. We excluded 15 patients because of inadequate histopathologic data or follow-up, 2 patients because of lack of postcontrast sequences, and 1 patient because of profound artifacts from an inserted maxillary prosthesis.

Image Acquisition

Images were acquired with a 1.5T scanner (Ingenia, Philips Healthcare) using a 16-channel neurovascular coil. Imaging started from the skull base to the thoracic inlet, with FOV anterior to posterior 230 mm, and 4-mm section thickness with no gap. Axial single-shot SE EPI DWI was acquired (b = 0 and 800 seconds/mm², TR/TE 2920/67 ms, section thickness 4 mm, and voxel size 3.2 × 2.6 × 4.0 mm). ADC was automatically generated by the implemented software, including precontrast axial SE T1WI (TE/TR: 21/633 ms), coronal SE T1 (TE/TR: 14/555 ms), axial SE T2WI (TE/TR: 10/7039 ms), sagittal SE T2WI (TE/TR: 100/3196.7 ms), and coronal T2 STIR (TR/TI =3500/150, TE =80 ms). A gadolinium-based contrast agent (dimeglumine gadopentetate) (0.1 mmol/kg) was administered using an injector with a flow rate of 2–3 mL/s followed by postcontrast T1WI with fat saturation, in the axial (TR/TE: 611/21 ms), sagittal (TR/TE:570/14 ms), and coronal (TR/TE:570/14 ms) planes.

Image Analysis and Applied Diagnostic Criteria

Data were transferred to an IntelliSpace Portal 9.0 workstation (Philips Healthcare). Scans were interpreted by 2 neuroradiologists with different levels of experience (5 and 15 years) in head and neck imaging, both of whom have been trained on NI-RADS reporting at our institution. NI-RADS has been the reporting template in use for posttreatment of HNSCC cases since early 2018 according to the ACR NI-RADS lexicon and the directions published in the Aiken and Hudgins⁵ guide article and the white paper of the ACR NI-RADS committee.¹ Readers had access to patients' demographic data (names and identification numbers had been anonymized), the therapeutic history, the pretreatment scans, and any other preceding posttreatment CECT with or without PET scan and were blinded to the clinical or pathologic outcome. Discrepant interpretations were resolved by consensus.

Three polygonal ROIs were inserted manually by the neuroradiologist with 5 years' experience on b = 800 images either on the same or consecutive axial sections, according to the size and extent of the primary targets and copied automatically on the corresponding ADC maps. ROIs were placed on the most enhancing parts in the presence of enhancement, excluding necrotic areas guided by contrast-enhanced-T1WI and T2WI in the same axial sections. The average ADC_mean value for the 3 measurements was estimated for all targets.

Two reporting template forms were used to evaluate primary targets, including the primary tumor site and any regional non-nodal target as associated perineural tumor spread; 1 was template confined to the existing NI-RADS lexicon and scoring system,^1,5 and the final score was provided according to the higher numeric scoring lesion in case of more than 1 primary target. The other template included the diffusion criteria in qualitative and quantitative forms and T2 SI of primary targets, and scoring was performed according to our modified rules to the current NI-RADS lexicon.

Based on the available literature^6,8 and the authors' experience, a defined scale for T2 SI evaluation was set: 1) isointense SI to the normal-appearing surrounding tissues, 2) very dark T2 SI (much lower than original tumor and muscle) (Figs 1A and 2A), 3) intermediate T2 SI (similar to the initial tumor signal), (Fig 1A), and 4) high T2 SI for tissues displaying SI between muscle SI and fluid, or much higher than the original tumor. Correlation with the pretreatment tumor signal was performed for 56 cases with available pretreatment CEMRI.

• Download figure
• Open in new tab
• Download powerpoint

FIG 1.

Concordance between NI-RADS category, T2 SI, and DWIs is demonstrated in this posttreatment MR imaging surveillance scan done after surgery and RTH for sinonasal squamous cell carcinoma. Coronal T2WI (A) and coronal contrast-enhanced T1WI (B) show a surgical bed discrete nodule (black arrow) that displays intermediate T2 SI (A) and postcontrast enhancement (B), fulfilling NI-RADS 3 and confirmed to be squamous cell carcinoma by histopathology. C. Axial ADC map of the surgical bed shows a corresponding low-ADC signal (black arrow). White arrows point to circumferential soft tissue thickening with sheetlike enhancement (B), which shows low T2 SI (A), absent diffusion restriction (D), and was confirmed to be posttreatment fibrous tissue.

For diffusion criteria, the scale used for visual interpretation of the signal to determine whether it was high or low was compared with the brain stem signal in ADC maps as follows: 1) diffusion restriction for low signal in the ADC map compared with the brain stem signal with high signal in DWI corresponding to abnormal enhancement (Fig 1C) and/or similar to the original tumor signal; and 2) facilitated diffusion for high signal in the ADC map compared with the brain stem with high or low signal on DWI (Figs 1D and 2C) or higher than the original tumor signal.

Image Analysis Using T2 SI and DWI as Modified Rules for NI-RADS Categories

We incorporated diffusion features and T2 SI into NI-RADS as modified rules. The presence of both diffusion restriction and intermediate T2 SI would upgrade the NI-RADS categories by 1 grade (from categories 1 to 2 and 2 to 3). Conversely, the absence of both would downgrade NI-RADS categories by 1 grade (Fig 2) (from categories 3 to 2 and 2 to 1). The presence of only 1 feature without the other feature would not alter the original NI-RADS risk category.

• Download figure
• Open in new tab
• Download powerpoint

FIG 2.

The first post-CRTH MR imaging follow-up for a known case of nasopharyngeal carcinoma with perineural tumor spread (PNTS) showing discordant findings between NI-RADS category, T2 SI, and DWI. There is a clear primary tumor site at the nasopharynx (not demonstrated here), yet a regional non-nodal target was noted. A, Axial T2WI shows orbital apex dark T2 SI tissue keeping with fibrotic scarring (arrow). B, This lesion show discrete postcontrast enhancement (arrow) (secondary to PNTS along the ophthalmic nerve) categorized as NI-RADS 3. C, Axial ADC map of the surgical bed shows a corresponding facilitated diffusion (arrow). According to our proposed modifying rules, downgrading to category 2 was done. Follow-up by PET/CT showed no FDG uptake (not demonstrated here) that was confirmed to be post-RTH fibrotic scarring by further follow-up. Note the right temporal lobe after RTH injury.

Standard of Reference

Criteria for negative disease included negative clinical and MR imaging follow-up performed after 12 months or negative biopsy results. Criteria for LTR included positive biopsy results, evident tumor on clinical examination, or evidence of disease progression on subsequent follow-up scans as determined by the Response Evaluation Criteria in Solid Tumors.¹²

Ethics Approval

All procedures performed in the study were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments. For this type of study informed consent was not required and waived by the ethical committee.

Statistical Analysis

The collected data were revised, coded, and tabulated using SPSS 20 (IBM). For descriptive statistics, the mean, SD, and range were used for numerical data; frequency and percentage were used for non-numeric data. For analytical statistics, the Student t test was used to assess the statistical significance of the difference in ADC_mean values between LTR and benign lesions. ROC curve analysis of ADC_mean values was performed according to the standard of reference. The diagnostic performance of ACR NI-RADS, DWI in its qualitative and quantitative forms, T2 SI, and NI-RADS combined with T2 SI and DWI were assessed in terms of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. Kappa statistics were used to compute the measure of agreement between 2 investigational methods, with values of 0.2–0.4 indicating fair agreement; 0.41–0.6, moderate agreement; 0.61–0.8, substantial agreement; and 0.8–1.0, almost perfect agreement. P < .05 was considered significant.