BACKGROUND.
Understanding of prognostic factors in parotid carcinoma has grown considerably. In particular, clinical tumor staging and histologic classification have been found to be prognostically important. Univariate and multivariate analyses have indicated that other variables, such as age, pain, skin invasion, and facial nerve impairment, are important predictors as well. In an actual patient, some of these factors are present and others are absent. However, a clinical tool incorporating this information, resulting in an individualized prognosis based on the combined effects of present adverse prognostic factors, has never been devised.
METHODS.
Of a cohort of 168 patients, 151 were evaluated to assess the prognostic value of clinical and pathologic factors in a multivariate proportional hazards analysis. Follow-up ranged from 1 to 278 months (median, 37 months). The end point was tumor recurrence. Identified prognostic factors and their hazard ratios were combined into prognostic scores.
RESULTS. Clinical
T classification, clinical N classification, pain, age at diagnosis, skin invasion, facial nerve dysfunction, perineural growth, and positive surgical margins acted as major factors predicting recurrence. A prognostic score (PS), generated by the weighted combination of the factors present in the individual patient, placed the patient in one of four subgroups with markedly different
prognoses. In the subgroups based on the preoperative prognostic score, 5-year recurrence free percentages ranged from 92% (in the group PS1Ο1) to 23% (in PS1Ο4). In the subgroups based on the postoperative prognostic score, which took into account the histologic details of the resected specimen, 5-year recurrence free percentages ranged from 95% (in the group PS2Ο1) to 42% (in PS2Ο4).
CONCLUSIONS.
The proposed subgrouping, which is based on the combined effects of key prognostic preoperative and postoperative factors, provides a practical prognostic grouping system for the clinician treating patients with parotid carci-