FDG-PET/CT Scanning a Reliable Imaging Surveillance Technique in LAHNSCC

Tim Van den Wyngaert, MD, PhD, Department of Nuclear Medicine Antwerp University Hospital

Tim Van den Wyngaert, MD, PhD

¹⁸Fluorodeoxyglucose-positron emission tomography/computed tomography (FDG-PET/CT) scanning reliably detected residual neck disease in newly diagnosed patients with locoregionally advanced head-and-neck squamous cell carcinoma (LAHNSCC).

Overall accuracy was 86.4% (95% CI, 79.3-91.3), according to findings from the prospective multicenter ECLYPS study published online in the Journal of Clinical Oncology.

The investigators reported that that the negative predictive value of FDG-PET/CT scanning performed 12 weeks after concurrent chemoradiotherapy (CCRT) was 92.1% (95% CI, 86.9-95.3). Sensitivity was 65.2% (95% CI, 44.9-81.2)—lower than researchers expected—specificity was 91.2% (95% CI, 84.1-95.3), and positive predictive value was 62.5% (95% CI, 45.5-76.9).

However, timing played a major role in sensitivity. A time-dependent analysis assessing the diagnostic performance of PET/CT scanning according to the time horizon of clinical follow-up after scanning showed sensitivity declining from 83.3% at 3 months to 59.7% at 12 months.

Researchers found no statistically significant difference in the overall diagnostic performance between central review using the Hopkins criteria and local nonstandardized assessment (AUROC, 0.78 vs 0.73; P = .336). Scans using the Hopkins criteria were significantly less likely to suggest diagnostic uncertainty, defined as a score of 3 (1.6% vs 10.4%; 95% CI of the difference, 2.6-15.0; McNemar exact P = .003). When 2 or more readers were involved in local assessment, the Hopkins criteria outperformed local interpretation (AUROC, 0.89 vs 0.70; P = .030), driven by a significant improvement in positive predictive value (33.3% vs 63.6%; P = .018)

“PET/CT can identify residual disease in patients who relapse up to a 9-month horizon after imaging with high sensitivity, but it is less able to do so for patients in whom residual disease was detected up to 12 months after imaging (sensitivity, 59.7%),” first author Tim Van den Wyngaert, MD, PhD, Department of Nuclear Medicine Antwerp University Hospital, Belgium, and coinvestigators wrote.

“This important finding may explain some of the heterogeneity in existing literature because of highly variable follow-up times. The strikingly low Hopkins scores in false-negative scans suggest that the disease was either microscopic and below the detection limit of PET/CT or was not metabolically active. Therefore, clinicians may consider a second surveillance scan at approximately 12 months after the end of CCRT, particularly in HPV-negative patients because they have the highest probability of residual disease after CCRT,” added Van den Wyngaert et al.

For ECLYPSE, 152 patients were recruited at 8 hospitals from March 2011 to February 2014. Twenty-five patients (16.4%) dropped out, mostly due to persistent or progressive primary tumor, distant metastases, or death before the imaging time point. A total of 125 patients were evaluable for efficacy.

Median duration of follow-up after completing CCRT was 20.4 months (interquartile range [IQR], 12.7 months). There were 17 deaths (13.6%), for an estimated 2-year overall survival probability of 85.9% (95% CI, 77.0-91.5). The 2-year locoregional control rate was 79.2% (95% CI, 70.3-85.7).

Of 125 patients with primary tumor control up to the 12-week imaging time point, 23 (18.4%) had residual neck disease during subsequent follow-up (95% CI, 12.0-26.3), meeting the expected rate of 20% (P = .654). Twenty-one patients (16.8%) underwent neck dissection.

The largest residual lymph node per patient as measured on the CT part of the 12-week PET/CT study had median short axis diameters of in Hopkins-negative patients 6 mm (IQR, 3 mm) and 8 mm (IQR, 3.5 mm) in Hopkins-positive. The proportion of misclassified among patients with a negative Hopkins neck score increased when residual lymph node size criteria or the presence of necrosis were applied. There was no significant improvement in misclassification rate with the addition of lymph node size or necrosis in patients with a positive Hopkins neck score, though the small size for that population scores (n = 24) made it difficult to draw conclusions from that observation.

Van den Wyngaert T, Helsen N, Carp L, et al. Fluorodeoxyglucose-positron emission tomography/computed tomography after concurrent chemoradiotherapy in locally advanced head-and-neck squamous cell cancer: The ECLYPS study [published online August 30, 2017]. J Clin Oncol. doi: 10.1200/JCO.2017.73.5845.