The median follow-up duration after radiotherapy was 36.7 months (range 14.0-90.5). A total of 17 patients experienced recurrences and two patients died. The patterns of first failure and overall failure at the time of the last follow-up are shown in Figure . Of the five patients who experienced regional recurrence as an isolated first recurrence, two experienced subsequent distant metastasis and two experienced subsequent local and distant failure. Of the four patients who experienced local recurrence as an isolated first recurrence, one patient experienced subsequent regional failure and one experienced subsequent regional and distant failure. Overall local, regional, and distant failure developed in 10, 11, and 10 patients, respectively. The 3-year OS, RFS, LC, RC, and DC rates were 95.7%, 62.7%, 77.2%, 77.1%, and 78.2%, respectively.
Figure 1 Patterns of (A) first failure and (B) overall failure. Of the five patients who experienced regional recurrence as an isolated first recurrence, two experienced subsequent distant metastasis and two experienced subsequent local and distant failure. Of (more ...)
Regional recurrence developed in 11 patients (crude rate 22%) at a median of 4.9 months after radiotherapy (range 2.3-23.6, interquartile range 2.6-7.0). Ipsilateral level II was the most common site of regional failure. Neck dissection for suspicious persistent or recurrent disease was performed in 12 patients at 2.3-24.5 months (median 6 months) after radiotherapy. Of these, two patients had no metastatic lymph nodes and 10 had metastatic lymph nodes in the neck. Of these 10 patients, seven were successfully salvaged, and the ultimate neck control rate was 92%. With the exception of two patients who developed distant metastasis, the other five of these seven patients were alive without disease at the last follow-up. Two patients with isolated local failure and one patient with isolated lung metastasis were also successfully salvaged with surgery and chemotherapy. Thus, 8 of 17 patients with recurrence were alive without disease at the last follow-up (median 24.6 months, range 9.7-88.1 months after recurrence).
We evaluated candidate parameters such as age, gender, primary site, T/N classification, initial lymph node size, radiotherapy modality, chemotherapy administration sequence, primary tumor response, post-radiotherapy lymph node size/necrosis, and PET reading of positive residual nodal disease to identify prognostic factors for regional control, and the results are listed in Table . On univariate analysis, initial nodal size > 2 cm (3-year RC 88.0% vs. 63.6%, p = 0.022), post-radiotherapy primary tumor response less than complete response (88.1% vs. 41.7%, p < 0.001), post-radiotherapy nodal size > 1.5 cm (83.5% vs. 20.0%, p < 0.001), and post-radiotherapy nodal necrosis (87.8% vs. 33.3%, p = 0.002) were associated with poor regional control. Administration of chemotherapy was heterogeneous and we divided patients into 3 groups; neoadjuvant plus concurrent, concurrent plus adjuvant, and concurrent only; corresponding 3-year RC rates were 71.4%, 94.1%, and 64.3%, respectively (p = 0.123). On multivariate analysis, a less-than-complete primary site response (hazard ratio 8.926, 95% confidence interval 2.38-33.47, p = 0.001) and post-radiotherapy nodal necrosis (hazard ratio 7.413, 95% confidence interval 2.03-27.14, p = 0.002) were identified as independent prognostic factors for regional control.
Univariate analysis to identify prognostic factors for regional control
PET as a first evaluation was performed in 31 patients. Only four of the 31 patients showed positive PET results with mSUV of 1.3, 2.8, 3.3, and 5.0, respectively. Two of them (mSUV 3.3 & 5.0) experienced regional failure. One patient had 1.7 cm necrotic lymph node with abnormal FDG uptake (mSUV 3.3) at 3 weeks post-radiotherapy. We decided that post-radiotherapy 3 weeks was too early to determine overall response and re-checked CT 8 weeks later. At 11 weeks, lymph node further decreased but was progressed in next follow-up CT. Another patient with regional recurrence had 1-cm lymph nodes without necrosis at 8 weeks after radiotherapy, and PET showed small but hypermetabolic lymph nodes (mSUV 4.2 and 5.0) at that time. Because there were small lymph nodes without necrosis, and FDG uptake decreased compared with pre-treatment value (mSUV 9.0), we observed without immediate neck dissection. However after 3 months, lymph nodes increased with necrosis and FDG uptake also increased up to mSUV 12.5, and the patients underwent salvage neck dissection. Since PET was performed in only a subset of the patients (n = 31) at the first follow-up, we analyzed subset analysis in patients who checked PET. Positive and negative PET results were reported in 4 and 27 patients, respectively, and the corresponding 3-year RC rates were 50% and 78.1% (p = 0.115). When we included patients who did not perform PET and divided patients into 3 groups (residual disease, normal/reactive change, and no PET), and 3-year RC rates were not significantly different (p = 0.349). We could not find any prognostic significance for regional control by the statistical analysis using several arbitrary cut off points of mSUVs regardless of the interpretation by the specialists in Nuclear Medicine.