Proceedings of the 52nd Annual ASTRO Meeting S563

Conclusions: The role of SBRT in the medically operable patient is yet to be defined. Our model indicates that SBRT may offercomparable overall survival and quality adjusted life expectancy as compared to surgical resection. Well powered prospective stud-ies comparing surgery versus SBRT in early lung cancer are warranted to further investigate the relative survival, quality of life andcost characteristics of both treatment paradigms.

Author Disclosure: A.V. Louie, None; G. Rodrigues, None; M. Hannouf, None; G.S. Zaric, None; D.A. Palma, None; J.Q. Cao,None; B.P. Yaremko, None; R. Malthaner, None; J.D. Mocanu, None.

2804 Reporting of Disease-specific Treatment Effects in Randomized Controlled Trials: Systematic Review and

Analysis of the Medical Literature

S. K. Lau, B. S. Rose, L. K. Mell

University of California San Diego, San Diego, CA

Purpose/Objective(s): Randomized controlled trials (RCTs) of cancer therapies frequently use composite primary endpoints thatcombine both disease-specific and non-specific events. This can lead to erroneous conclusions about treatment effects unless effectson the primary endpoint are partitioned into disease-specific and non-specific components. However, the extent to which treatmenteffects are partitioned in contemporary RCTs is unknown.

Materials/Methods: We reviewed 470 RCTs published between 2006 and 2008 in 8 leading medical journals: NEJM, JAMA, Lan-cet, JNCI, JCO, Lancet Oncology, IJROBP, and Radiother Oncol. Search criteria included ‘‘cancer,’’ ‘‘leukemia’’ or ‘‘lymphoma’’in the article title. We excluded RCTs conducted in patients with recurrent/metastatic disease (173), RCTs not published in print (65),secondary analyses (36), prevention studies (21), and RCTs using toxicity, quality of life, or response rate primary endpoints (89). Ofthe remaining 116 RCTs, 27 (23%) analyzed a disease-specific primary endpoint, and 14 (12%) did not identify the primary endpointof the study; analysis was conducted on the remaining 75 RCTs, all of which used an event-free survival (EFS) primary endpoint.Each article was scored on a scale from 1-4 for completeness in analyzing disease-specific and non-specific components of thetreatment effect on the primary endpoint, by 2 independent reviewers. Scoring discrepancies between two reviewers (minor)were resolved by a third independent reviewer; residual scoring discrepancies (major) were resolved by consensus.

Results: Of the 75 RCTs using an EFS primary endpoint, 35 (47%) used overall survival (OS), 21 (28%) used disease-free survival(DFS), and 19 (25%) used an alternative EFS endpoint. Minor and major scoring discrepancies occurred in 21 (28%) and 5 (7%) ofRCTs, respectively. The kappa statistic for scoring agreement for the initial review was 0.66, indicating substantial scoring agree-ment. 28 RCTs (37%) reported both disease-specific and non-specific partitioning of the treatment effect (score = 4), 21 (28%)reported incomplete partitioning of the treatment effect (score = 3), and 26 (34%) did not partition the treatment effect (score =2), including 7 RCTs that reported no disease-specific effects (score = 1). Specious disease-specific inferences based on EFSendpoints were identified in 2 RCTs. The mean score for breast cancer RCTs was significantly higher (3.34) than non-breast cancerRCTs (2.75) (p = 0.015). No significant difference in scores was observed according to journal.

Conclusions: Incomplete partitioning of treatment effects on primary endpoints is common in contemporary RCTs. Greaterattention to partitioning disease-specific and non-specific treatment effects in RCTs is needed.

Author Disclosure: S.K. Lau, None; B.S. Rose, None; L.K. Mell, None.

2805 Weekly Self-reported Tobacco use and Cessation during Definitive Radiotherapy or Chemoradiotherapy

for Head and Neck Cancer

G. W. Warren1, S. Arnold2, J. Valentino2, J. Sims2, M. Reid1, M. Cummings1, J. Marshall1, A. Martella1, T. Gal2, M. Kudrimoti2

1Roswell Park Cancer Institute, Buffalo, NY, 2University of Kentucky, Lexington, KY

Purpose/Objective(s): Though evidence suggests that positive self-reported tobacco use (SRTU) in clinical cancer populations isassociated with a poor therapeutic outcome, most SRTU assessments are performed only at initial diagnosis/consult. There isvirtually no data on longitudinal SRTU during cancer treatment. This study reports baseline and weekly SRTU in a cancer treatmentpopulation.

Materials/Methods: An IRB approved study at the University of Kentucky was initiated to assess the relationship betweennicotine exposure and therapeutic outcome in non-metastatic head and neck cancer patients treated with definitive radiotherapy(RT) or chemoradiotherapy (CRT). This interim analysis reports weekly SRTU during RT or CRT.

Results: Of 50 patients accrued to date, 80% are male, median age is 56 (range, 39-75), 92% are white, 92% are overall stage III-IVB, and 84% were treated with definitive CRT. Median smoking history was 30 pack years (range, 3-138). Baseline SRTUassessment at study entry (initial consultation) demonstrates 82% report any history of SRTU and 40% report current SRTU. Over-all, 30% of patients changed SRTU status between diagnosis/initial consult and week 1 of RT or CRT. In patients who report neg-ative SRTU at baseline, 20% convert to positive SRTU by week 1 (maximum conversion 27% at week 3 of treatment). In patientswho report positive SRTU at baseline, 55% convert to negative SRTU by week 1 of treatment. Secondhand smoke exposure atweek 1 was reported in 33% of baseline negative SRTU patients and 30% of baseline SRTU positive patients. Overall, positiveSRTU during the final week of treatment was reported in 18% of patients. At baseline, 12% of patients (2 negative SRTU and4 positive SRTU) reported pharmacologic cessation efforts and 8% of patients reported pharmacologic cessation efforts at anytime during treatment. Two patients (both reporting current SRTU) were hospitalized and were not able to complete treatment.

Conclusions: Data demonstrate a large (30%) transition in SRTU status among cancer patients from the time of initial diagnosis/consult and the time of cancer treatment. Data further demonstrate a significant proportion of cancer treatment patients are exposedto secondhand smoke. Given the potentially adverse effects of tobacco exposure on therapeutic outcome, substantial effort shouldbe made to assess SRTU both at the time of diagnosis/initial consult and routinely throughout cancer treatment. This significantlyimpacts the design of clinical practice and research protocols that only assess SRTU prior to the start of cancer treatment.

Author Disclosure: G.W. Warren, None; S. Arnold, None; J. Valentino, None; J. Sims, None; M. Reid, None; M. Cummings,None; J. Marshall, None; A. Martella, None; T. Gal, None; M. Kudrimoti, None.