2018-19 Section 7-Neoplastic and Inflammatory Diseases of the Head and Neck eBook

Oral Oncology 79 (2018) 9–14

C.L. Barney et al.

Impact of locoregional failure on distant failure

Table 1 Patient characteristics.

On univariate analysis LRF predicted for subsequent distant failure (p = < .001). The results from Table 2 guided the construction of Cox proportional hazards models for multivariate analysis using distant failure as an end point and the independent prognostic values of age > 60, N-stage (ICON-S 2-3), T-stage ( ≥ T3/T4), use of concurrent cetuximab, and LRF were tested. In this model, LRF trended toward an increased risk of a subsequent distant failure (HR 2.10, 95% CI 0.95 – 4.62; p = .07) with the only signi fi cant covariate being use of concurrent cetuximab (HR 2.44, 95%CI 1.16 – 5.12; p = .02). When in- cluding only pretreatment factors in the multivariate model, LRF was the only signi fi cant covariate (HR 3.10, 95%CI 1.49 – 6.45; p = .002). A separate analysis of only patients who received concurrent ce- tuximab was performed to identify prognostic and predictive factors in this subgroup. The 3-year LRC for patients receiving AFRT was 83.0% compared to 63.0% with SFRT (p = .05; Fig. 3 ). On multivariate ana- lysis, the only factor which remained associated with decreased LRC was the use of SFRT (HR 0.29, 95% CI 0.09 – 1.02; p = .05). Similarly, N-stage (ICON-S 2-3) was the only factor associated with decreased DMFS (HR 0.34, 95% CI 0.13 – 0.87; p = .03), RFS (HR 0.50, 95% CI 0.25 – 1.00; p = .05), and OS (HR 0.36, 95% CI 0.14 – 0.94; p = .04). To our knowledge this is the largest single institution study com- paring clinical outcomes in cisplatin versus cetuximab CRT in p16+ OPC. This is also the fi rst study incorporating the newly established ICON-S/AJCC 8th edition nodal staging. In our analysis, concurrent cisplatin was associated with LRC, DMFS, RFS, and OS compared to cetuximab. Moreover, these results were obtained in relatively well- balanced groups; when combining those who received cetuximab through randomization on clinical trial (32%) or by patient choice (17.6%), we estimate that at least 50% of patients who received ce- tuximab were cisplatin eligible. Riaz et al. reported similar results, however, their study was limited by sample size (n =62, with only 23 receiving cetuximab) and popu- lation heterogeneity bias (27% had non-oropharyngeal primary tumors, which was prognostic on multivariate analysis) [9] . Another recent prospective study showed a trend toward improved LRC and OS with the use of cisplatin compared to cetuximab in p16+ OPC, however, the small number of patients (n =33) did not allow for signi fi cance to be reached for any of the outcomes analyzed [10] . Our study focuses only on p16+ oropharyngeal primary tumors and bene fi ts from a much larger sample size. For example, our cetuximab cohort alone had 68 patients in which RT fractionation schedules happened to be well di- chotomized. This allowed for a meaningful subgroup analysis, which not only identi fi es important prognostic factors in cetuximab CRT but also provides unique insight into the impact of di ff erent RT fractiona- tion schedules in this population. Furthermore, most ongoing rando- mized trials in this population have strati fi ed patients by RPA with the prospect of isolating a subgroup in which concurrent cetuximab may be non-inferior to cisplatin. Our data is novel in that it is the fi rst to report that improved survival with the use of concurrent cisplatin compared to cetuximab is independent of RPA grouping. Emphasizing this point is that even our IR-RPA cisplatin patients had numerically better 3-year OS compared to the LR-RPA cetuximab patients (84.8% vs. 80.3%; p = .21). Not all prior studies have reported such poor outcomes with the use of cetuximab CRT in p16+ OPC. In post hoc analysis of the IMCL-9815 trial, Rosenthal et al. reported higher 3-year rates of LRC, RFS, and OS (87.0, 82.1, and 87.8%, respectively) compared to those reported in our study [5] . Reasons for this discrepancy in outcomes may be related to Cetuximab subgroup analysis Discussion

Characteristic

Drug

p-value

Cisplatin (n = 137)

Cetuximab (n = 68)

Male, n (%)

117 (85.4) 57 (22 – 80) 42 (7 – 78)

60 (88.2) 61 (42 – 79) 27 (4 – 65)

.58 .01

Age (y), median (range) Follow-up (mo), median (range)

< .001

Smoking (> 10 pack-years), n (%)

58 (42.3)

32 (47.8)

.46

T-Stage, n (%)

.81

1 2 3 4

23 (16.8) 65 (47.4) 26 (19.0) 23 (16.8) 2 (1.5) 10 (7.3) 14 (10.2) 54 (39.4) 42 (30.7) 15 (10.9) 57 (41.6)

13 (19.1) 29 (42.6) 16 (23.5) 10 (14.7) 1 (1.5) 6 (8.8) 9 (13.2) 23 (33.8) 25 (36.8)

N-Stage (AJCC 7th Ed.), n (%)

.75

N0 N1

N2a N2b N2c

N3

4 (5.9)

N-Stage 2 – 3 (ICON-S), n (%)

29 (42.6)

.89

ECOG-PS, n (%)

.25

0 1

86 (62.8) 51 (37.2)

37 (54.4) 31 (45.6)

RPA, n (%)

.87

Low-risk

85 (62.0) 52 (38.0) 19 (13.9) 14 (10.2)

43 (63.2) 25 (36.8) 8 (11.8) 21 (31.3)

Intermediate-risk

Postoperative CRT, n (%) Altered Fractionation, n (%)

.68

< .001

(p = .01; Table 3 ), while advanced T-stage remained associated with decreased RFS (p = .03; Table 3 ). When stratifying by RPA, 3-year OS was 97.1% in LR-RPA patients who received cisplatin and 80.3% in those who received cetuximab (p < .001; Fig. 2 ). Similarly for IR-RPA patients, the 3-year OS was greater with concurrent cisplatin than ce- tuximab (84.8 vs 53.5%, respectively; p < .001; Fig. 2 ). Abbreviations : AJCC=American Joint Committee on Cancer; ICON-S = International Collaboration on Oropharyngeal cancer Network for Staging; ECOG PS = Eastern Cooperative Oncology Group performance status; CRT= chemoradiation. At the time of analysis, 10 (7.3%) patients who received concurrent cisplatin had experienced a locoregional failure (LRF): 5 (3.6%) at the primary site, 2 (1.5%) in the neck, and 3 (2.2%) in both the neck and primary site. Meanwhile 19 (27.9%) cetuximab patients had experi- enced a LRF: 6 (8.8%) at the primary site, 9 (13.2%) in the neck, and 4 (5.9%) in both the neck and primary site. The 3-year LRC for patients who received concurrent cisplatin was 92.7% and 65.4% for those who received cetuximab ( Fig. 1 C). Distant failure was experienced by 15 (10.9%) patients who had received cisplatin and 19 (27.9%) who had received cetuximab. The 3-year DMFS was 88.3% and 71.2% in cis- platin and cetuximab patients, respectively ( Fig. 1 D). There were no signi fi cant di ff erences in LRC (p = .360) or DMFS (p = .230) between those treated with high-dose versus weekly cisplatin ( Supplemental fi gure ). Concurrent cetuximab was associated with decreased LRC (p < .001) and DMFS (p < .001) on univariate analysis ( Table 2 ). On multivariate analysis ( Table 3 ) these associations persisted for LRC (HR 0.20, 95% CI 0.09 – 0.44; p < .001) and DMFS (HR 0.33, 95% CI 0.16 – 0.64; p = .001). Other signi fi cant covariates included advanced T- stage for LRC (p = .04) and advanced N-stage for DMFS (p = .03; Table 3 ). Locoregional control and distant metastasis

95