BackgroundDue to the occult anatomic location of the nasopharynx and frequent presence of adenoid hyperplasia, the positive rate for malignancy identification during biopsy is low, thus leading to delayed or missed diagnosis for nasopharyngeal malignancies upon initial attempt. Here, we aimed to develop an artificial intelligence tool to detect nasopharyngeal malignancies under endoscopic examination based on deep learning.MethodsAn endoscopic images-based nasopharyngeal malignancy detection model (eNPM-DM) consisting of a fully convolutional network based on the inception architecture was developed and fine-tuned using separate training and validation sets for both classification and segmentation. Briefly, a total of 28,966 qualified images were collected. Among these images, 27,536 biopsy-proven images from 7951 individuals obtained from January 1st, 2008, to December 31st, 2016, were split into the training, validation and test sets at a ratio of 7:1:2 using simple randomization. Additionally, 1430 images obtained from January 1st, 2017, to March 31st, 2017, were used as a prospective test set to compare the performance of the established model against oncologist evaluation. The dice similarity coefficient (DSC) was used to evaluate the efficiency of eNPM-DM in automatic segmentation of malignant area from the background of nasopharyngeal endoscopic images, by comparing automatic segmentation with manual segmentation performed by the experts.ResultsAll images were histopathologically confirmed, and included 5713 (19.7%) normal control, 19,107 (66.0%) nasopharyngeal carcinoma (NPC), 335 (1.2%) NPC and 3811 (13.2%) benign diseases. The eNPM-DM attained an overall accuracy of 88.7% (95% confidence interval (CI) 87.8%–89.5%) in detecting malignancies in the test set. In the prospective comparison phase, eNPM-DM outperformed the experts: the overall accuracy was 88.0% (95% CI 86.1%–89.6%) vs. 80.5% (95% CI 77.0%–84.0%). The eNPM-DM required less time (40 s vs. 110.0 ± 5.8 min) and exhibited encouraging performance in automatic segmentation of nasopharyngeal malignant area from the background, with an average DSC of 0.78 ± 0.24 and 0.75 ± 0.26 in the test and prospective test sets, respectively.ConclusionsThe eNPM-DM outperformed oncologist evaluation in diagnostic classification of nasopharyngeal mass into benign versus malignant, and realized automatic segmentation of malignant area from the background of nasopharyngeal endoscopic images.

No abstract

No abstract

added to concurrent chemoradiotherapy significantly improves survival for patients with locoregionally advanced nasopharyngeal carcinoma, but the optimal induction regimen remains unclear.OBJECTIVE To determine whether induction chemotherapy with paclitaxel, cisplatin, and capecitabine (TPC) improves survival vs cisplatin and fluorouracil (PF) prior to chemoradiotherapy for patients with stage IVA to IVB nasopharyngeal carcinoma. DESIGN, SETTING, AND PARTICIPANTSThis randomized, open-label, phase 3 clinical trial recruited 238 patients at 4 hospitals in China from October 20, 2016, to August 29, 2019. Patients were 18 to 65 years of age with treatment-naive, nonkeratinizing stage IVA to IVB nasopharyngeal carcinoma and an Eastern Cooperative Oncology Group performance status of 0 to 1.INTERVENTIONS Patients were randomly assigned (1:1) to receive induction chemotherapy with two 21-day cycles of TPC (intravenous paclitaxel [150 mg/m 2 , day 1], intravenous cisplatin [60 mg/m 2 , day 1], and oral capecitabine [1000 mg/m 2 orally twice daily, days 1-14]) or PF (intravenous cisplatin [100 mg/m 2 , day 1] and fluorouracil [800 mg/m 2 daily, days 1-5]), followed by chemoradiotherapy. MAIN OUTCOMES AND MEASURESThe primary end point was failure-free survival in the intention-to-treat population. Secondary end points included distant metastasis-free survival, locoregional relapse-free survival, overall survival, tumor response, and safety. RESULTS Overall, 238 eligible patients (187 men [78.6%]; median age, 45 years [range, 18-65 years]) were randomly assigned to receive TPC (n = 118) or PF (n = 120). The median follow-up duration was 48.4 months (IQR,. Failure-free survival at 3 years was 83.5% (95% CI, 77.0%-90.6%) in the TPC group and 68.9% (95% CI, 61.1%-77.8%) in the PF group (stratified hazard ratio [HR] for recurrence or death, 0.47; 95% CI, 0.28-0.79; P = .004). Induction with the TPC regimen resulted in a significant reduction in the risk of distant metastases (stratified HR, 0.49 [95% CI, 0.24-0.98]; P = .04) and locoregional recurrence (stratified HR, 0.40 [95% CI, 0.18-0.93]; P = .03) compared with the PF regimen. However, there was no effect on early overall survival (stratified HR, 0.45 [95% CI, 0.17-1.18]; P = .10). The incidences of grade 3 to 4 acute adverse events and late-onset toxicities were 57.6% (n = 68) and 13.6% (16 of 118), respectively, in the TPC group and 65.8% (n = 79) and 17.9% (21 of 117), respectively, in the PF group. One treatment-related death occurred in the PF group.CONCLUSIONS AND RELEVANCE This randomized clinical trial found that induction chemotherapy with 2 cycles of TPC for patients with stage IVA to IVB nasopharyngeal carcinoma improved failure-free survival compared with 2 cycles of PF, with no increase in the toxicity profile.

IMPORTANCE Capecitabine maintenance therapy improves survival outcomes in various cancer types, but data are limited on the efficacy and safety of capecitabine maintenance therapy in metastatic nasopharyngeal carcinoma (NPC).OBJECTIVE To investigate the efficacy and safety of capecitabine maintenance therapy in metastatic NPC. DESIGN, SETTING, AND PARTICIPANTSThis randomized phase 3 clinical trial was conducted at Sun Yat-sen University Cancer Center from May 16, 2015, to January 9, 2020, among 104 patients with newly diagnosed metastatic NPC who had achieved disease control after 4 to 6 cycles of induction chemotherapy with paclitaxel, cisplatin, and capecitabine. The final follow-up date was May 30, 2021. All efficacy analyses were conducted in the intention-to-treat population.INTERVENTIONS Eligible patients were randomly assigned (1:1) to receive either capecitabine maintenance therapy (1000 mg/m 2 orally twice daily on days 1-14) every 3 weeks plus best supportive care (BSC) (capecitabine maintenance group) or BSC alone after 4 to 6 cycles of induction chemotherapy.MAIN OUTCOMES AND MEASURES Progression-free survival (PFS). Secondary end points were objective response rate, duration of response, overall survival, and safety. RESULTSThis study included 104 patients (84 men [80.8%]; median age, 47 years [IQR, 38-54 years]), with 52 assigned to the capecitabine maintenance group and 52 assigned to the BSC group. After a median follow-up of 33.8 months (IQR, 22.9-50.7 months), there were 23 events (44.2%) of progression or death in the capecitabine maintenance group and 37 events (71.2%) of progression or death in the BSC group. Median PFS survival was significantly higher in the capecitabine maintenance group (35.9 months [95% CI, 20.5 months-not reached]) than in the BSC group (8.2 months [95% CI, 6.4-10.0 months]), with a hazard ratio of 0.44 (95% CI, 0.26-0.74; P = .002). Higher objective response rates and longer median duration of response were observed in the capecitabine maintenance group (25.0%; 40.0 months) compared with the BSC group (objective response rate, 25.0% [n = 13] vs 11.5% [n = 6]; and median duration of response, 40.0 months [95% CI, not reached-not reached] vs 13.2 months [95% CI, 9.9-16.5 months]). The most common grade 3 or 4 adverse events during maintenance therapy were anemia (6 of 50 [12.0%]), hand-foot syndrome (5 of 50 [10.0%]), nausea and vomiting (3 of 50 [6.0%]), fatigue (2 of 50 [4.0%]), and mucositis (2 of 50 [4.0%]). No deaths in the maintenance group were deemed treatment-related. CONCLUSIONS AND RELEVANCEIn this phase 3 randomized clinical trial, capecitabine maintenance therapy significantly improved PFS for patients with newly diagnosed metastatic NPC who achieved disease control after capecitabine-containing induction chemotherapy. Capecitabine exhibited manageable toxic effects.