After obtaining approval from the Johns Hopkins Institutional Review Board, the prospectively collected pathology database at Johns Hopkins Hospital (JHH), Baltimore, MD (September 2, 1986 to May 19, 2007), was searched for pancreatic carcinoma specimens reported to contain any amount of squamous differentiation. The search identified 45 patients with a diagnosis of adenocarcinoma with squamous differentiation. These 45 cases represented 1.2% of a total of 3651 patients who underwent pancreatic resection (pancreaticoduodenectomy [PD], distal or total pancreatectomy) for malignant or benign conditions during that same period. Excluded from final analysis were 2 patients found to have distant metastasis at surgical resection, 3 patients with very low survival months (<2 months), and 2 other patients who were lost to follow-up. The final study includes 38 patients for whom archival histopathologic material and clinical outcome data could be obtained.
Clinical data including demographics, operative details, adjuvant treatment, as well as outcome were collected. Data on preoperative characteristics including symptoms (eg, weight loss, abdominal pain, and jaundice), tobacco use, and comorbid conditions (eg, diabetes mellitus, hypertension, myocardial infarction, coronary artery disease, and chronic obstructive pulmonary disease) were collected. Other collected variables included intraoperative blood loss, units of blood transfused, length of hospital stay, and postoperative complications.
All available histologic slides were re-reviewed by a single pathologist (J. Davison). The proportion of squamous differentiation in the primary carcinoma was assessed using morphological criteria (including absence of gland lumen formation, mucin production, cell shape, quality of intercellular borders and cytoplasm, and the presence of keratinization). In certain cases, it was difficult to quantify the percentage of the carcinoma with squamous or glandular differentiation. Squamous differentiation most often occurred in poorly differentiated carcinomas, and the adenocarcinoma and squamous carcinoma components were usually admixed within the same section of the tumor. In such cases, a poorly differentiated component was determined to be either glandular (adenocarcinoma) or squamous based on whether glandular or squamous differentiation predominated in that region of the carcinoma that was sampled. Each section of the primary tumor was scored separately, and an overall estimate was based on the average, taking into account the amount of tumor on each slide.
The presence or absence of an “undifferentiated” component was also noted. Undifferentiated carcinoma was defined as a malignant epithelial neoplasm lacking identifiable histologic features of squamous or glandular differentiation [8
]. The undifferentiated components of these neoplasms typically exhibited sarcomatoid features or were composed of sheets of discohesive, highly anaplastic cells. The presence or absence of perineural invasion, vascular invasion, and lymph node metastases were documented, as was whether the involvement was due to infiltration by the squamous component, glandular component, or both. An average of 10 slides (median, 9; range, 2-40) of carcinoma (including metastases) was reviewed for each case. Information about tumor size, the status of margins, as well as total lymph node counts was obtained from the original pathology reports and confirmed by review of the available slides. The current American Joint Committee on Cancer (AJCC) pathologic tumor stage was determined from slide review and a review of the original pathology report. Cases in which the uncinate (retroperitoneal) margin was determined to be positive were categorized as T4 lesions under the presumption that the tumor involved the superior mesenteric artery.
Details regarding adjuvant chemoradiation (CRT) were retrospectively collected. For patients with available chemotherapy detail, chemotherapeutic agents included 5-fluorouracil (5-FU), gemcitabine, and capecitabine. Patients who underwent CRT at JHH received 5-FU–based chemoradiation as previously described [2
]. For patients with available radiotherapy detail, the median radiation dose was 5040 Gy (range, 4500-5000 Gy).
2.1. Statistical analysis
Survival was determined and verified using Cancer Center abstracting services and the Social Security Death Index. Statistical analyses were performed using JMP statistical software, version 8 (SAS Institute Inc., Cary, N.C.) [12
]. Tests of differences between variables were performed using t
tests, F tests, and χ2
tests. For characteristics with individuals missing data, χ2
tests were performed including only those with known status, as indicated. The primary outcome variable was median overall survival (OS) defined as the time from surgical resection of PASC to death. If the patient was still alive, the date of last follow-up was used as censored data. Survival curves were estimated using Kaplan-Meier techniques [13
]. The proportion of individuals surviving up to 1, 2, and 5 years was calculated using life tables. Univariate analyses were performed on variables known to be prognostically important for pancreatic cancer as previously described [9