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

Oral Oncology 73 (2017) 152–159

K.Y. Zhan et al.

Table 3 Subgroup analysis of AJCC 8th edition pathologic stages I-III.

Stage I (n = 3001)

Stage II (n = 663)

Stage III (n = 78)

p

Mean age (SD) at diagnosis

57 (9)

58 (9)

58 (9)

Sex

0.001

Male

2508 (83.6%) 493 (16.4%)

590 (89%) 73 (11%)

69 (88.5%)

< 10 a

Female

Race

0.033

White Black Other

2844 (95.7%)

613 (93.6%)

74 (94.9%)

< 10 a < 10 a

80 (2.7%) 48 (1.6%)

33 (5%)

<10 a

Radiotherapy

<0.001

< 10 a

No Yes

619 (20.6%) 2367 (78.9%)

89 (13.4%) 570 (86%)

72 (92.3%)

<10 a

< 10 a

Unknown

15 (0.5%)

Chemotherapy

<0.001

No Yes

1465 (48.8%) 1476 (49.2%)

202 (30.5%) 444 (67%)

12 (15.4%) 64 (82.1%)

<10 a

< 10 a

Unknown

60 (2%)

Subsite

<0.001

Base of tongue

863 (28.8%) 2039 (67.9%)

211 (31.8%) 407 (61.4%)

23 (29.5%) 51 (65.4%)

Tonsil

<10 a

< 10 a < 10 a

Soft palate

12 (0.4%) 87 (2.9%)

Other

43 (6.5%)

a Small cell counts hidden per NCDB policy; SD = standard deviation.

3745 patients — the largest study to date on this topic. While new pa- thologic parameters such as pN-staging and number of lymph nodes involved were studied in a multinational fi ve-center study, they had not yet been validated in a national registry [8] . The NCDB is a much more representative national sample — collecting data from both academic and non-academic/research hospital settings — and has recently been used to successfully support the adopted changes in the clinical staging system for HPV+ OPSCC [12] . Key points in our results include: con- fi rmation of the dramatic downstaging of pathologic stage IV to stage I disease from AJCC 7th to AJCC 8th; the excellent hazard discrimination of AJCC 8th overall pathologic staging ( Fig. 2 ); the equivocal survival curves of pN0 vs. pN1 (1 – 4 nodes involved) ( Fig. 3 ); and evidence that ENE could play a prognostic role in HPV+ OPSCC. Groome et al. previously outlined four criteria for an ideal staging scheme: hazard consistency (similar survival rates for patients within any given stage group); hazard discrimination (di ff erent survival rates between groups); outcome prediction; and balanced patient distribution across stage groups (maximizing statistical power in each group) [13] . In AJCC 7th edition, comparisons between HPV+ OPSCC and HPV − OPSCC were problematic in hazard discrimination and outcome pre- diction, as Stage IV HPV+ patients often had better survival than early- stage HPV − disease. HPV+ OPSCC ’ s rising incidence compounded this problem (and skewed stage distribution towards advanced disease), making it di ffi cult to accurately convey prognosis to patients, families, and other clinicians. Characteristics of our study cohort and staging distribution closely matches recent literature on HPV+ OPSCC: predominantly younger Caucasian males with AJCC 7th stage IV disease [12] . The poor hazard discrimination and prognostication of AJCC 7th edition for HPV+ OPSCC ( Fig. 2 a) is clear and resulted in the impetus for recent changes [3] . The dramatic shift from AJCC 7th Stage IV to AJCC 8th Stage I in our cohort is consistent with previous, multicenter trials that con- tributed to 8th edition changes [3,4,8] . Haughey et al. ’ s 2016 fi ve- center trial is the largest series prior to our analysis, with 704 surgi- cally-treated HPV+ OPSCC patients. They had a distribution of 70% Stage I, 24% stage II, 6% stage III. Adjuvant treatment was given in 79%

of cases (compared to 81% in our NCDB analysis). Five-year OS was 90%, 84%, and 48%, respectively [8] . The excellent hazard dis- crimination of AJCC 8th edition overall staging is clearly evident ( Fig. 2 b) and now provides meaningful and accurate outcome predic- tion (2nd and 3rd criteria suggested by Groome et al. [13] ). These improvements are critical for developing treatment de-escalation trials and will likely result in much-needed management guidelines speci fi c to HPV+ disease [6,9] . Nonetheless, additional strati fi cation should be sought within the large Stage I subgroup for numerical balancing, sta- tistical power, and prognostic clarity [13] . The change from nodal size/side to nodal count in AJCC 8th edition is based on several surgical studies that found regional metastatic nodal count — not size or laterality — to be more predictive of survival in HPV + OPSCC [8,14 – 18] . While our pN2 cohort had signi fi cantly worse survival than pN0-1, pN0 and pN1 survival curves were similar (4-year OS of 88% and 91% respectively, p = 0.065). Haughey et al. only found ≥ 5 nodes (pN2) to be associated with signi fi cantly worse survival in their study, as N0 patients had similar survival outcomes as those with 1 – 4 positive nodes. However, it cannot be ignored that T-stage dis- tribution and management of pN0 and pN1 were di ff erent in our cohort, as triple modality therapy was used in 52.2% of pN1 cases vs. 15.4% in pN0. Moreover, radiotherapy was utilized in only 40% of pN0 vs. 86% in pN1. T2-T4 disease was also more common in pN0 than pN1. A combination of higher T-stages, less adjuvant treatment, and other patient factors could explain the equivocal survival between pN0 and pN1 in our analysis. These same imbalances make our fi ndings re- garding pN2 disease even more compelling; despite increased multi- modality treatment and lower T-stage tumors, pN2 patients still had worse prognosis. Additional research with lymph node count and other pathologic variables in patient- and treatment-matched studies is needed. A potential area of additional staging strati fi cation is the presence of pathologic ENE. We report a modest decrease in 4-year OS when ENE is present (92% vs. 85%, p < 0.001). When strati fi ed by N-stage ( Fig. 4 ), ENE retained a signi fi cant impact on survival for pN1 (N = 2469, p = 0.004), but did not for pN2 (N = 428, p = 0.061). An et al.

179