most patients had multiple similar episodes. Nadir SpO 2 averaged 758%. The mean longest duration of saturation below 90% was 21min 24sec15min. Discussion/Conclusion: Severe and pro- longed arterial hypoxemia was a consistent finding in our surgical population, despite aggressive preoperative diagnosis and treat- ment of OSA as well as postoperative use of CPAP beginning in recovery. Although we expected some degree of postoperative hy- poventilation and obstructive sleep apnea, we were amazed to discover the degree and frequency of desaturation, when the data was unblinded. No patient experienced cardiopulmonary arrest, or even a significant alteration in vital signs, in spite of severe and repeated arterial hypoxemia. In no instance, were significant hy- poxemic episodes even suspected. Supplemental O 2 likely blunts the hypoxemic response to hypoventilation and V/Q mismatch and perhaps to OSA episodes; however, O 2 supplementation will pre- vent diagnosis and specific therapy.1,2 Past experience has shown O 2 therapy will not prevent narcotic-induced respiratory depres- sion and occasional associated cardio-respiratory arrest.3 Supple- mental O 2 will render pulse oximetry useless as a monitor of ventilation.4 Pulse oximetry, with an alarm set for saturation less than 90%, for ten seconds, would have detected 100% of significant hypoxemic episodes, thus permitting appropriate diagnosis and therapy. We believe we have merely uncovered the tip of an iceberg. Some patients do arrest, postoperatively. Often, it is assumed that narcotic-induced respiratory depression is the etio- logic factor, albeit rare. Our recommendation is to routinely and appropriately monitor every patient, in order prevent that occurrence. References: 1. Downs J. Has oxygen administration delayed appropriate respiratory care? Fallacies regarding oxygen therapy. Respir Care 2003;48:611–20. 2. Downs JB, Smith RA. In- creased inspired oxygen concentration may delay diagnosis and treatment of significant deterioration in pulmonary function. Crit Care Med 1999;12:2844 – 6. 3. The pulse oximeter as a monitor of ventilation. Anesthesia Patient Safety Foundation Newsletter 2006; 21(4):61–7. 4. Fu ES, Downs JB, Schweiger JW, Miguel RV, Smith FA. Supplemental oxygen impairs detection of hypoventilation by pulse oximetry. Chest 2004;126:1552– 8. QS263. INCIDENCE OF FOLLICULAR THYROID CARCINO- MA: A 17 YEAR EXPERIENCE OF PRE-OPERATIVE FINE NEEDLE ASPIRATION CLASSIFICATION AS FOLLICULAR LESION VERSUS FOLLICULAR NEO- PLASM. James W. Suliburk 1 , Michelle D. Williams 2 , Gregg A. Staerkel 2 , Naifa L. Busaidy 2 , Jeff E. Lee 2 , Doug B. Evans 2 , Nancy D. Perrier 2 ; 1 University of Texas Medical School at Houston, Houston, TX; 2 University of Texas MD Anderson Cancer Center, Houston, TX Background: Fine needle aspiration (FNA) classification of follicu- lar thyroid nodules remains a challenging problem as the determi- nation of benign from malignant disease is rarely possible. While research efforts search for Methods to pre-operatively classify follic- ular nodules that harbor malignancy, the decision for operative excision is largely based on clinical concerns, size of the nodule and age of the patient. We chose to re-examine our clinical outcomes based on the cytopathologic classification of follicular nodules at our institution. The purpose of this study was to assess the incidence of follicular carcinoma in thyroid nodules where the FNA diagnosis was follicular lesion versus follicular neoplasm and evaluate the utility of these diagnostic categories in triaging patients to sugery. We hy- pothesized that an interpretation of follicular neoplasm would carry a higher incidence of follicular carcinoma. Methods: We conducted a retrospective review of all patients with a diagnosis of follicular lesion or follicular neoplasm by FNA treated during the last 17 years at a single institution. Each record was independently reviewed. The diagnosis of follicular lesion was made on FNA when some colloid or variability in nuclei was suggestive of but not diagnostic for an adenomatous process. Follicular neoplasm was diagnosed on FNA when monotonous follicular cells with a paucity of colloid were present for which a follicular adenoma or carcinoma was favored. All cases with a suspicion for papillary thyroid carcinoma were excluded. All cases with a malignant diagnosis on the surgical resection spec- imen were evaluated for correlation with FNA site of biopsy and pathologic findings. Chi-squared analysis was used to determine significant differences in incidence of follicular carcinoma between follicular lesion and follicular neoplasm groups. A P- value less than 0.05 was considered significant. Results: Overall 614 patients with thyroid nodules underwent 672 FNAs: 518 were diagnosed as follic- ular lesion and 154 as follicular neoplasm. 356 (58.0%) patients underwent surgical resection composed of 239 (46.1%) of the patients with a preoperative diagnosis of follicular lesion and 117 (75.9%) of the patients with a cytopathologic diagnosis of follicular neoplasm. Follicular carcinoma was present in 16 (6.7%) of the follicular lesion group and 18 (15.4%) of the follicular neoplasm group. Overall, the incidence of malignancy including papillary thyroid carcinoma- follicular variant, papillary thyroid carcinoma and follicular carci- noma was 12.6% (30 of 239) for follicular lesions and 22.2% (26 of 117) for follicular neoplasms by FNA classification. The incidence of follicular carcinoma as well as overall malignancy was significantly higher for the follicular neoplasm group. Conclusions: At our insti- tution, the cytopathologic evaluation by FNA of thyroid follicular nodules and their classification into follicular lesions versus follicu- lar neoplasms serves to identify patients at greater risk for follicular carcinoma. Additionally, other malignancies are identified in a greater proportion of follicular neoplasms. Although malignancy may be present in either follicular neoplasms or follicular lesions, follicular neoplasms represent a higher risk nodule and thus this information is helpful when weighing the risk benefit ratio of surgi- cal excision in a patient with a follicular nodule. QS264. FACTORS THAT PREDICT MALIGNANCY IN HU¨ RTHLE CELL NEOPLASMS OF THE THYROID. Yi Wei Zhang, David Y. Greenblatt, Daniel Repplinger, Anna Bargren, Joel Adler, Herbert Chen; University of Wisconsin, madison, WI Background: Hürthle cell neoplasms (HCN) of the thyroid gland account for 3% to 10% of all thyroid tumors. Hürthle cell carcinoma (HCC) is more aggressive than other well-differentiated thyroid can- cers such as follicular and papillary thyroid carcinoma, with higher rates of recurrence and metastasis. Many surgeons recommend total or near-total thyroidectomy for the treatment of HCC, while Hürthle cell adenomas (HCA) are adequately treated with thyroid lobectomy. Discriminating between HCA and HCC either before or during sur- gery is difficult. Currently, HCA and HCC can only be distinguished by permanent histology, as neither fine-needle aspiration biopsy nor frozen-section analysis is reliable for differentiation. The purpose of this study was to determine if any pre-operative or intra-operative factors are predictive of malignancy in patients with HCN. Meth- ods: Between May 1994 and January 2007, 1199 patients underwent thyroid surgery at an academic medical center. Medical records from 52 consecutive patients who had pre-operative diagnosis of HCN underwent thyroid resections were reviewed. Statistical analysis was performed to identify patient and tumor factors that were pre- dictive of malignancy. Results: Of the 52 patients with HCN, 44 had adenomas and 8 had carcinomas on final pathologic examination, yielding a malignancy rate of 15.4%. Patients with HCC were signif- icantly older than those with HCA (70  5 years vs. 53  2 years, p  0.002). Patients with carcinoma also had significantly larger thy- roid nodules (4.5  0.7 cm vs. 2.5  0.2 cm, p  0.001). None of the resected Hürthle cell neoplasms less than 2 cm in diameter were found to be malignant. Furthermore, the prevalence of malignancy increased with nodule size: 17% of nodules measuring 2cm to 4cm, and 40% of those larger than 4cm were determined to be HCC. Out of the eight patients with HCC, only one had recurrence of the disease (12.5%). After up to 13 years of follow-up, there was no 371 ASSOCIATION FOR ACADEMIC SURGERY AND SOCIETY OF UNIVERSITY SURGEONS—ABSTRACTS