1

Introduction

Many important factors require analysis to adequately assess risk factors for disease outcome in head and neck squamous cell carcinoma (HNSCC) . Perineural invasion (PNI) is acknowledged to be one of these important markers. In noncutaneous HNSCC, histologic evidence of PNI, or “neurotropic carcinomatous spread,” is acknowledged to be a poor prognostic factor and indicative of the need for adjuvant therapy . Although first described in the 19th century, it was the early works of Ballantune et al and Batsakis in the 20th century that brought this feature to greater clinical awareness. Batsakis’s broad definition of PNI in 1985 characterized it as tumor cell invasion in, around, and through nerves—a broad category that encompasses most observations. In their review, Liebig et al advocate PNI definition as “… the finding of tumor cells within any of the 3 layers [epi-, peri-, or endoneurium] of the nerve sheath….”

Although great interest has been shown in developing molecular markers for PNI , such work remains preliminary, with markers lacking specificity, sensitivity, and predictive value . Because the mere presence of PNI has important prognostic significance , we hypothesized that objectively quantifying its extent may further enhance our understanding of clinical outcomes. The aim of the current study was to determine if the extent of PNI in relation to tumor margins is predictive of disease-free survival. To that end, we define a novel classification system of PNI subcategories as intratumoral (IT), peripheral, or extratumoral (ET) and examine the prognostic value of this system as it relates to clinical outcome.

2

Methods

A retrospective review of 142 patients with HNSCC who underwent primary surgical treatment at the University of California, Los Angeles, Medical Center from January 2004 through December 2007 was performed. Inclusion criteria were biopsy-proven HNSCC (all subtypes) without previous treatment, and all head and neck sites and levels of differentiation were included. The study period was chosen to allow sufficient time for analysis of patient outcomes. Clinical information collected included patient age, sex, alcohol and tobacco use, tumor location, TNM stage, postoperative adjuvant chemotherapy and/or radiation treatment, and patient outcome. For each patient, a thorough histopathologic examination of the surgical excision specimen stained with hematoxylin and eosin was performed to document all foci of PNI on each slide if present. For each focus of PNI identified, the distance to the tumor edge (in millimeters) was measured, with the tumor edge being considered 0 mm. Hence, a negative value indicated an IT location, and a positive value indicated an ET location. Moreover, a peripheral location was arbitrarily defined as being −0.2 to 0 mm from the tumor edge. In addition, the size of the largest nerve involved (in millimeters) regardless of its location with respect to the tumor edge was measured. The density of PNI was calculated for each case, and the number of PNI foci and number of tumor sections were examined. Perineural invasion was subcategorized according to its greatest PNI extent, that is, IT, peripheral, or ET. Perineural invasion was examined by a pathologist (C.L.) who was blinded to patient clinical outcomes.

Of note, 27 cases were recategorized from PNI negative to PNI positive (6 cases) and from PNI positive to PNI negative (21 cases). This was based on more stringent criteria for defining PNI. The maximum extent of PNI was determined by recording the most positive value of the PNI location assigned to each case. In some cases, the most positive value was a negative value, corresponding to an IT focus of PNI.

Cox regression analysis was used to assess the relationship between the maximum extent of PNI with the time to regional disease recurrence. The relationship between PNI categories and the number of positive nodes was examined using a Poisson regression model. Poisson regression is used when the outcome, in this case, the number of nodes, is a count-type variable. The relationships between PNI status and categorical clinical variables (eg, T stage) were examined using χ ² tests. Kaplan-Meier plots were constructed for time to recurrence using the PNI subcategories. The log-rank test was used to compare the recurrence times between the PNI subcategories. Subjects with missing PNI measures were included as a separate PNI category for completeness.

2

Methods

A retrospective review of 142 patients with HNSCC who underwent primary surgical treatment at the University of California, Los Angeles, Medical Center from January 2004 through December 2007 was performed. Inclusion criteria were biopsy-proven HNSCC (all subtypes) without previous treatment, and all head and neck sites and levels of differentiation were included. The study period was chosen to allow sufficient time for analysis of patient outcomes. Clinical information collected included patient age, sex, alcohol and tobacco use, tumor location, TNM stage, postoperative adjuvant chemotherapy and/or radiation treatment, and patient outcome. For each patient, a thorough histopathologic examination of the surgical excision specimen stained with hematoxylin and eosin was performed to document all foci of PNI on each slide if present. For each focus of PNI identified, the distance to the tumor edge (in millimeters) was measured, with the tumor edge being considered 0 mm. Hence, a negative value indicated an IT location, and a positive value indicated an ET location. Moreover, a peripheral location was arbitrarily defined as being −0.2 to 0 mm from the tumor edge. In addition, the size of the largest nerve involved (in millimeters) regardless of its location with respect to the tumor edge was measured. The density of PNI was calculated for each case, and the number of PNI foci and number of tumor sections were examined. Perineural invasion was subcategorized according to its greatest PNI extent, that is, IT, peripheral, or ET. Perineural invasion was examined by a pathologist (C.L.) who was blinded to patient clinical outcomes.

Of note, 27 cases were recategorized from PNI negative to PNI positive (6 cases) and from PNI positive to PNI negative (21 cases). This was based on more stringent criteria for defining PNI. The maximum extent of PNI was determined by recording the most positive value of the PNI location assigned to each case. In some cases, the most positive value was a negative value, corresponding to an IT focus of PNI.

Cox regression analysis was used to assess the relationship between the maximum extent of PNI with the time to regional disease recurrence. The relationship between PNI categories and the number of positive nodes was examined using a Poisson regression model. Poisson regression is used when the outcome, in this case, the number of nodes, is a count-type variable. The relationships between PNI status and categorical clinical variables (eg, T stage) were examined using χ ² tests. Kaplan-Meier plots were constructed for time to recurrence using the PNI subcategories. The log-rank test was used to compare the recurrence times between the PNI subcategories. Subjects with missing PNI measures were included as a separate PNI category for completeness.