2559 Background: Pegylated recombinant human megakaryocyte growth and development factor (PEG-rHu-MGDF aka MGDF) and recombinant granulocyte colony stimulating factor (G-CSF) promote the maturation of hematopoietic progenitor cells. Healthy volunteers/donors have received MGDF in phase I/II clinical trials and G-CSF in allogeneic peripheral blood stem cell transplantation procedures. Herein, we review clinical findings for five previously healthy volunteers/donors who developed hematologic malignancies after the use of MGDF or G-CSF. Methods: Clinical information related to hematologic malignancies were reviewed for three volunteers who had participated in a phase I/II clinical trial with MGDF and two donors who underwent G-CSF mobilized peripheral blood stem cell harvesting procedures for sibling allogeneic stem cell transplantation for acute leukemia. Results: Mantle cell, diffuse large B-cell lymphoma, and chronic lymphocytic leukemia were diagnosed three to five years after exposure among three volunteers who received MGDF. For one of these patients, autoimmune thrombocytopenia and antibodies to MGDF that cross-reacted with endogenous thrombopoietin had developed shortly after MGDF administration and persisted until lymphoma chemotherapy was administered. Following chemotherapy, all three achieved complete remission, although one patient subsequently relapsed. Acute myelogenous leukemia was diagnosed four to five years after exposure in two donors who underwent G-CSF primed stem cell harvests prior to their siblings’ allogeneic stem cell transplantation. Following intensive chemotherapy, one of these patients died from acute leukemia and the second is now in complete remission. Conclusion: Controversy exists over the appropriateness of administering hematopoietic growth factors to healthy individuals. While a causal relationship with hematologic malignancies is uncertain, long-term follow-up among healthy individuals who receive hematopoietic growth factors is needed. No significant financial relationships to disclose.

14621 Background: After undergoing definitive treatment for a primary localized disease, prostate cancer patients may experience a rise in their prostate specific antigen (PSA) levels. Treating PSA failure with hormonal treatment has many health related quality of life (HRQOL) implications, including urinary, bowel, sexual, and male hormonal problems. Predictors of choice between hormonal treatment versus watchful waiting have not been investigated. Methods: Patients were approached after consecutive rises in PSA levels (n = 31). Patients completed HRQOL and decision satisfaction questionnaires, and a literacy assessment. Results: Patients were between 56 and 85 years old; 55% were African American. 71% of African Americans and 50% of whites had low functional literacy. 58% of patients chose hormonal therapy to treat their PSA rise; 81% of patients reported urinary problems. All patients reported decision satisfaction (see Table). Factors associated with castration versus watchful waiting were primarily related to poor urologic function, and were not specifically prostate cancer related (dysuria, nocturia, urination frequency). Conclusions: Primary treatment of urinary dysfunction, rather than castration, should be evaluated as initial therapy for prostate cancer patients with PSA failure. [Table: see text] No significant financial relationships to disclose.

6074 Background: Half of cancer therapies are used off-label, but regulations prohibit package inserts from describing toxicities that occur only in these settings. In 2003, RADAR reported VTE rates of 20% to 62% when thalidomide, approved for leprosy treatment, was used off-label (with dexamethasone) for multiple myeloma. In 2005, the Connecticut attorney general requested that a black box warn of high rates of VTE when thalidomide is used off-label. Subsequently, the thalidomide analogue, lenalidomide, received FDA approval as a myelodysplasia therapy. In phase II/III trials, multiple myeloma response rates were 60% to 92% with lenalidomide/dex therapy. Methods: Published literature, abstracts, and package inserts were reviewed for VTE rates in lenalidomide-treated multiple myeloma. Results: VTE rates were 8.5% to 75% in multiple myeloma treated with lenalidomide and dex or erythropoietin; rates were <3.4% when aspirin prophylaxis was added. The FDA approved package insert for lenalidomide preemptively includes a black box warning describing high VTE rates with off-label use of lenalidomide and advises physicians that VTE prophylaxis should be considered, although the effectiveness of various prophylaxis regimens is unknown. Conclusions: Six months after the attorney general requested that a black box warning describe high VTE rates with off-label use of thalidomide, the manufacturer preemptively included an analogous black box warning in the package insert for the thalidomide analogue, lenalidomide, but did not include similar warnings for this toxicity in the package insert for thalidomide. The FDA should require that package inserts describe severe toxicities that commonly occur with off-label use of cancer therapies. This is especially important for severe class-related toxicities such as lenalidomide- and thalidomide- associated VTE. [Table: see text] No significant financial relationships to disclose.

14541 Background: Studies have shown that African American populations have low rates of surgery, while similar assessments based on literacy skills have not been reported. Methods: Newly diagnosed prostate cancer patients (n = 311) were approached at 3 hospitals: private, county, and VA. Patients completed demographic and quality of life questionnaires as well as a literacy assessment at baseline, 3 month, and 12 month intervals. Results: Majority (60%) of African American patients were at low functional literacy. In multinomial regression analysis controlling for patient age, clinical stage at presentation, comorbidity, and treatment site, both African American race (RRR 2.4, 95% CI 1.35–5.91) and low literacy (6th grade reading level or below) were significantly and independently associated with a greater likelihood to receive external beam radiation treatment only in relation to other treatment options (AOR 4.26, 95% CI 2.34–7.75). Conclusions: In our inner city population of prostate cancer patients, white patients with high literacy skills opted for radical prostatectomy and persons with low literacy and/or African American race opted for radiation therapy. No significant financial relationships to disclose.

6062 Background: RADAR, a NIH funded R01 project, obtains reports of serious ADRs by conducting hypothesis-driven active surveillance efforts in hematology/oncology. The FDA and pharmaceutical sponsors conduct passive surveillance efforts based on review of reports voluntarily submitted by health care workers. Methods: Completeness and timeliness of ADR reports and dissemination efforts by RADAR versus the FDA/pharmaceutical sponsors were compared. Results: Individual reports were more complete in RADAR databases with fewer total reports (341 versus 1,341). Pharmaceutical sponsors disseminated ADR data 1 year (median) earlier as revised package inserts (PIs), but RADAR publications were more complete. Conclusion: Compared to pharmacovigilance efforts by the FDA/pharmaceutical suppliers, RADAR is more complete but less timely. [Table: see text] No significant financial relationships to disclose.

6072 Background: Previous studies have found that patient treatment choice for localized prostate cancer is primarily dependent on the specialty of the treating physician. Herein, investigators with the Chicago Cohort Study of Prostate Cancer evaluated treatment choice at three multi-modality hospitals. Methods: Newly diagnosed prostate cancer patients (N=150) were approached at diagnosis at a VA, a county and a private hospital. Patients completed demographic and medical questionnaires and were followed to record treatment choice. Results: In a multivariate analysis, controlling for age and stage, hospital site was a significant predictor of treatment via radical prostatectomy at the VA site (AOR: 3.1, 96%CI: 1.4–6.9, p=0.006). However, age less than 65 was also a significant predictor of radical prostatectomy (AOR: 8.6, 95%CI: 1.8–41.5, p=0.008). Hospital site was also a significant predictor or external beam therapy as patients at Ingalls hospital were more likely to receive this treatment (AOR: 10.1, 95%CI: 3.0–33.5, p<0.001). Receiving therapy at the county hospital was also a significant predictor of receiving external beam therapy and hormonal therapy (AOR: 3.1, 95%CI: 1.5–6.5, p=0.002). Conclusions: At multiple modality sites, the specific hospital rather than the treating physician or specific patient characteristics appears to be a predictor of the type of treatment that a patient will receive. No significant financial relationships to disclose.

Background: At ASH 2004 we reported that prior to 2004, eight drugs had received accelerated approval (AA) for ten hematologic oncology indications, only one had converted to full approval, and three serious adverse events associated with two of these drugs were identified post-approval. Following our presentation, Congressman Ed Markey (D-Massachussets) in June 2005 issued a report (“Conspiracy of Silence”) that raised similar concerns about the low rates of conversion from accelerated to regular approval and the failure of the FDA in ensuring that drug companies complete the required post-marketing studies. We update the experience with accelerated FDA approval for hematologic oncology. Methods: Information was obtained from the FDA on new drug applications and supplements for new uses, package inserts, and medical literature reviews. Results: Of 11 drug indications that received AA for hematologic oncology between 1995 and 2005, AA was generally granted for 9 based on clinical trials with <100 patients that identified high response rates. In the past year, none have applied for conversion to regular approval. Bortezomib received regular approval for an indication different than that granted for AA. Three serious adverse drug reactions were identified for two indications (gemtuzumab ozogamicin, imatinib for GIST) within 2 years of AA approval. Conclusion: Since 2004, one new drug received AA for a hematologic oncology indication, one drug received regular approval for an indication different from that indicated under AA, and although 1 more year has elapsed, conversion has yet to occur for nine AA hematologic oncology indications. It is imperative that the FDA establish clear milestones regarding conversion to regular approval at the time AA is granted. Drug Name (AA Date; Years Lapsed Since AA) AA Indication Years to AA/Regular Approval N in Trial (Overall Response Rate) *Regular approval granted for myeloma failing one treatment Denileukin difitox (2/99; 6.5) T-cell lymphoma 1.17/- 71 (30%) Cytarabine liposomal (4/99; 6.3) Lymphomatous meningitis 0.46/- 14 (71%) Gemtuzumab ozogamicin (5/00; 5.35) CD33 positive AML 0.58/- 142 (30%) Alemtuzumab (5/01; 4.3) CLL 1.375/- 93 (33%) Imatinib mesylate (5/01) CML 0.25/2 864 (80%) Ibritumomab tuixetan (2/02; 3.5) Low-grade NHL 1.3/- 157(68%) Imatinib mesylate (2/02; 3.5) GIST 0.29/- 147 (54%) Imatinib mesylate (12/02; 2.7) CML peds 0.5/- 39 (67%) Bortezomib (5/03) Multiple myeloma w/ 2 prior chemo treatments* 0.46/2 193 (35%) Tositumomab (6/03/2.2) NHL 4.0/- 59 (71%) Clofarabine (12/04; 0.7) Relapsed/refractory ALL peds 0.75/- 49 (20%)

Since its introduction in 1988, recombinant human erythropoietin (epoetin) has been standard treatment for patients with anemia due to chronic kidney disease. From 1998 to 2004, nearly 200 epoetin-treated persons with chronic kidney disease developed antibodies to epoetin, resulting in pure red cell aplasia (PRCA). The majority of these patients received Eprex, an epoetin alfa product marketed exclusively outside the United States. Herein, we report on the long-term outcome of these individuals. For 170 chronic kidney disease patients who developed epoetin-associated PRCA and had 3 months or more follow-up information available, case reports from the Food and Drug Administration and epoetin manufacturers were reviewed for information on clinical characteristics of the patients, immunosuppressive treatments, epoetin responsiveness, and hematologic recovery. Overall, 64% of the PRCA patients received immunosuppressive therapy, including 19 who also underwent a renal transplantation. Thirty-seven percent experienced a hematologic recovery, with higher hematologic recovery rates among PRCA patients who received immunosuppressive therapy (57% vs 2%, P < .001). Among 34 patients who received epoetin after the onset of PRCA, 56% regained epoetin responsiveness. The highest rates of epoetin responsiveness were observed among persons whose antierythropoietin antibodies were undetectable when epoetin was administered (89%). Among chronic kidney disease patients with epoetin-associated PRCA, epoetin discontinuation and immunosuppressive therapy or renal transplantation is necessary for hematologic recovery. Reinitiation of epoetin therapy among individuals could be considered if antierythropoietin antibodies are undetectable.

CONTEXTIn 1998, a multidisciplinary team of investigators initiated RADAR (Research on Adverse Drug events And Reports), a clinically based postmarketing surveillance program that systematically investigates and disseminates information describing serious and previously unrecognized adverse drug and device reactions (ADRs).OBJECTIVETo describe the structure, operations, and preliminary findings from the RADAR project and related dissemination efforts by pharmaceutical suppliers and the US Food and Drug Administration (FDA).DESIGNAfter identifying a serious and unexpected clinical event suitable for further investigation, RADAR collaborators postulated clinical hypotheses and derived case series and incidence estimates from physician queries, published and unpublished clinical trials, published case reports, FDA databases, and manufacturer sales figures.RESULTSRADAR investigators identified 16 types of serious ADRs among 1699 patients, of whom 169 (10%) died as a result of the reaction. Initial cases were identified by 7 RADAR investigators, 4 collaborating physicians, 2 attorneys, and by reviewing 3 published reports. Additional sources included queries of occupational health programs and medical directors of interventional cardiology laboratories (3 types of ADRs), published manuscripts and clinical trials (11 types of ADRs), review of medical records at a RADAR site (2 types of ADRs), unpublished clinical trial reports (3 types of ADRs), and reports from attorneys, family members, or patients (4 types of ADRs). Incidence estimates, ranging from 0.4% to 33%, were derived from 5 clinical trial reports, 2 physician queries, and 2 observational databases. Laboratory support for hypotheses included identification of 3 neutralizing antibodies and 3 histopathological findings. ADR reports were disseminated as 8 revised package inserts, 7 "dear doctor" letters, and 9 peer-reviewed articles.CONCLUSIONA new, clinically based, hypothesis-driven approach to postmarketing surveillance may supplement existing regulatory surveillance systems and improve patient safety.

Background: With the recent reporting of 191 cases of epoetin induced PRCA (Bennett CL, NEJM 2004), information is needed about the long-term follow-up of individuals with this diagnosis. While 47 cases from France, Germany, and England have been reported recently (Verhelst D, Lancet 2004), these data are limited by short follow-up and absence of information from other regions.

6002 Background: The FDA adopted accelerated approval in 1992 in order to improve access to new drugs based on studies that evaluated markers likely to predict clinical benefit. Methods: The performance of this program was assessed by reviewing new oncology licensing applications and supplements for new uses approved by the FDA from 1995 to September 2003. Data included: the application filing date for each agent, the date of approval, the types of studies that were assessed during FDA review, and the endpoints used in these studies. Results: Of the 21 oncologic indications which received accelerated FDA approval, subsequent studies that were required for conversion to regular approval were submitted and accepted (n=5) [capecitabine, dexrazoxane, docetaxiel, imitinab mesylate, irinotecan], submitted for regular approval but rejected by the FDA (n=3), or have not been submitted (n=14). Surrogate outcomes included response rate (n=16), symptom improvement (n=1) and reduced time to recurrence (n=1). Fourteen approvals were based on results of phase II clinical trials, while only 5 were based on phase III studies. Fewer than 200 patients were evaluated for efficacy for about 2/3 of the indications. Post-marketing studies identified potentially fatal adverse drug reactions for four of the drugs, three of which were identified within 7 months of initial FDA approval (capecitabine-associated interaction with warfarin, imatinib-associated hemorrhage in GIST, and gemtuzumab-associated VOD). Conclusions: To date, drugs which have received accelerated FDA approval have been frequently evaluated in small, short-term phase II clinical trials, have rarely been converted to regular FDA approval, and can have potentially fatal adverse drug reactions identified in the first year of marketing. No significant financial relationships to disclose.

6129 Background: Measuring patient preferences for health states related to prostate cancer treatment outcomes provides valuable input to help patients make informed treatment decisions. The most accepted method for direct utility assessment is the standard reference gamble (SG). Methods: A pilot study was conducted with 21 newly diagnosed prostate cancer patients at a VA medical center. Each patient completed 6 standard gambles with health states including training exercises with mild and severe arthritis, current health, urinary incontinence, impotence, and both incontinence and impotence. They also rated each state using a visual analogue scale (VAS). The research assistant assessed patient comprehension. Responses were considered invalid if (1) they were inconsistent if combined symptoms were rated as better than either symptom alone, (2) the subject rated all scenarios at 0.5, (3) rank order of states differed from VAS, or (4) the subject rated all scenarios equal to perfect health AND the research assistant judged that there was incomprehension. Literacy was assessed in 19 of the subjects using the Rapid Estimate of Adult Literacy in Medicine (REALM). For comparative purposes, 5 urologists provided preferences for the same scenarios. Results: The mean REALM score (59) corresponded with 7th-8th grade literacy (low literacy is defined as < than 9th grade). Only 57% (12) patients gave potentially valid responses. There was a trend toward better performance in high literacy patients that did not reach statistical significance (χ2 p = 0.37). All urologists gave valid responses. Conclusions: Standard reference gamble assessments of utilities may be difficult to accurately assess in a VA population, particularly in patients with low literacy. Utilities directly elicited from low literacy patients should be done with caution, particularly when used to guide shared clinical decision-making or policy formation. No significant financial relationships to disclose.

Background: In 1992, the FDA extended the accelerated approval program from HIV and AIDs, to include drugs for other diseases that are serious or life-threatening, appear to provide benefit over available therapy, or for which no other therapy is available. Approval is based on a surrogate endpoint, and pharmaceutical companies are required to confirm safety and efficacy by conducting clinical trials with clinically relevant outcomes. Since 1995, eight drugs used to treat hematologic malignancies have received accelerated approval.

6616 Background: The incidence of GO-associated VOD and predisposing factors are uncertain. Barriers to understanding this toxicity are under- and incomplete ADR reporting. This concern is especially important for drugs such as GO, which receive accelerated FDA approval. Methods: We evaluated the completeness of GO-associated VOD ADR cases reported to the in the context of clinical trials (n=39) and non-clinical trial settings (n= 54), and reviewed incidence rates and predisposing factors reported in the medical literature. Results: clinical trial cases contained more information on abdominal examination findings (90% vs 59%), GO dose (92% vs 63%), GO schedule (92% vs 63%), VOD onset date (85% vs 51%), hepatic transaminases levels (72% vs 39%) and bilirubin (82% vs 57%) (p's<0.05). ADR reports indicated that GO-associated VOD was characterized by: elevated hepatic enzymes (82%), bilirubin (88%) and weight gain, ascites, or painful hepatomegaly (88%) occurring at a median of 10 days after GO administration (IQR 7.5–20). The medical literature was reviewed for incidence rates when GO was administered in both FDA-approved and "off-label" regimens, such as with hematopoietic stem cell transplantation (HSCT): Conclusions: 1) ADR reporting is more complete in the clinical trial versus the non-clinical trial setting; 2) VOD is a serious ADR that occurs at low rates when GO mono-therapy is used at the FDA approved dose and schedule and at higher rates when GO is administered at non-FDA approved doses or schedules, concurrently with other chemotherapeutic agents or when allogeneic HSCT is performed close to the time of GO administration; and 3) for cancer drugs such as GO which receive accelerated approval from the FDA, consideration should be given to mandating prospective post-marketing safety registries. No significant financial relationships to disclose.

BACKGROUND:Between 1988 and 1998, antibody-associated pure red-cell aplasia was reported in three patients who had undergone treatment with recombinant human erythropoietin (epoetin). Between 1998 and 2000, 13 such cases were reported from France--12 in patients who had received the Eprex formulation of epoetin alfa and 1 in a patient who had received Neorecormon (a formulation of epoetin beta); both are products that are marketed outside the United States.METHODS:We obtained reports of epoetin-associated pure red-cell aplasia from the Food and Drug Administration and from the manufacturers of Eprex, Epogen (another formulation of epoetin alfa), and Neorecormon. The numbers of case reports and estimates of exposure-adjusted incidence were analyzed according to the product, the cause of anemia, the route of administration, the country in which pure red-cell aplasia was identified, and the date on which pure red-cell aplasia was reported.RESULTS:Between January 1998 and April 2004, 175 cases of epoetin-associated pure red-cell aplasia were reported for Eprex, 11 cases for Neorecormon, and 5 cases for Epogen. Over half these cases had occurred in France, Canada, the United Kingdom, and Spain. Between 2001 and 2003, the estimated exposure-adjusted incidence was 18 cases per 100,000 patient-years for the Eprex formulation without human serum albumin, 6 per 100,000 patient-years for the Eprex formulation with human serum albumin, 1 case per 100,000 patient-years for Neorecormon, and 0.2 case per 100,000 patient-years for Epogen. After procedures were adopted to ensure appropriate storage, handling, and administration of Eprex to patients with chronic kidney disease, the exposure-adjusted incidence decreased by 83 percent worldwide.CONCLUSIONS:After the peak incidence of Eprex-associated pure red-cell aplasia was reached in 2001, interventions designed in response to drug-monitoring programs worldwide resulted in a reduction of more than 80 percent in the incidence of pure red-cell aplasia due to Eprex.

Background: Follow-up reports describing the long-term prognosis of 47 patients with epoetin induced PRCA have been confined to a single study from France, England, and Germany, investigated through active surveillance efforts. There is concern that reporting quality may differ between active versus voluntary reporting efforts.

6019 Background: African Americans present with a higher stage of disease for prostate cancer. Multivariate analysis indicates that low literacy rather than race, is the most likely predictor of advanced stage (Bennett, JCO 1998). We evaluate the relation of race and literacy with Gleason score, PSA level and treatment. Methods: Newly diagnosed prostate cancer patients were recruited from three Chicago Hospitals (N=143), including two equal access healthcare systems. Patients completed a demographic questionnaire and a literacy evaluation, using a 66 word standard list, the Rapid Estimate of Adult Literacy in Medicine (REALM). Stage of presentation and treatment decision were abstracted from medical records. Results: Patients were primarily African American (76%), of lower socioeconomic status (72%) with an income < $20,000, with low literacy skills (65% with <9th grade) and were older than 65 years of age (60%). Bivariate analysis indicated that race and low literacy were significantly associated with Gleason score and PSA respectively. In a multivariate analysis using literacy, race and age, low literacy was the only significant predictor of PSA > 10ng/mL (OR= 2.8, 95% CI: 1.1–7.7). Of literay, race and age, race was the only predictor for a high Gleason score (OR= 3.0, 95% CI: 1.2–7.1). Neither race nor literacy were significant predictors of treatment for younger or older patients. Conclusions: Low literacy, rather than African American race, is the most significant predictor of elevated PSA at presentation, while race is the most significant predictor of a high gleason score. In order to increase informed decision making about PSA screening, new strategies for health communication need to address the needs of men with lower literacy levels. No significant financial relationships to disclose.

6086 Background: Shared decision making may be difficult for older prostate cancer patients of lower SES. Although physicians and patients make shared treatment decisions about the option of radiation, surgery, or watchful waiting, functional and psychosocial obstacles may limit the effectiveness of this process among older men of lower socioeconomic status with newly diagnosed prostate cancer. Methods: A cohort (n= 85) of lower socioeconomic status White (25%) and African American patients (75%), 65 year and older with newly diagnosed prostate cancer were queried on social support and evaluated for health literacy, hearing, vision, activities of daily living, nutrition, depressive mood, and cognitive and mobility function using validated tools including the Rapid Estimate for Adult Literacy in Medicine, Mini Nutritional Assessment Short Form, and the Vulnerable Elderly Survey. Results: Subjects were mainly low SES (29% within the poverty range) and African American (74.2%). Mean age was 72 years (SD = 5), and one third lived alone. Impairments identified included: <7th grade health literacy (30%); cognitive impairment (32%); vulnerable (increased risk of decline in activities of daily living and death) (15%); mobility (19%); poor nutrition (22%); visual impairment (8%); hearing impairment (21%); depressive mood (10%); and lack of social support (4%). Overall, 73% of patients had at least one impairment. Conclusions: We conclude that among older, lower SES prostate cancer patients, three quarters have functional and psychosocial impairments that may limit the feasibility of shared decision making approaches. No significant financial relationships to disclose.

Background– Although thalidomide (Th) was granted FDA approval in 1998 for erythema nodosum leprosum, >99% of its use is off-label for cancer. As with all off-label therapies, FDA regulations prohibit dissemination of clinical information by the manufacturer. Nonetheless, concern over off-label toxicities arose in a 2001 phase III study in which 28% of myeloma patients developed deep vein thrombosis (DVT) and pulmonary embolism (PE) with concurrent Th-chemotherapy. We investigated the clinical characteristics and reporting quality of all adverse event (AE) reports describing Th-associated thromboembolism (TE). Methods– Data sources included 190 AE reports obtained from the FDA and 38 published phase II and III trials with 1,857 Th-treated individuals (for rate estimation), with exclusion of cases/trials reported in duplicate. AE reporting quality comparisons were made for events, which occurred in clinical trials (n=49) or in clinical practice settings (n=141). Results– Overall, 190 cancer patients were reported as having Th-associated TE, based on AE reports submitted to the drug manufacturer (n=49 from prospective clinical trials and n=131 from non-clinical trial setting) or directly to MedWatch (n=10). Only 6 cases were reported through the System for Th Education and Prescriber Safety (STEPS), an FDA mandated teratogenicity-focused safety program. DVT/PE occurred at a median of 52 days Th use (range, 6 to 469 days). Completeness and certainty of DVT/PE diagnoses were best for AE reports submitted directly to the manufacturer from clinical trials versus those from non-clinical trial settings (p<0.0001). Highest rates of TE were among patients who received concurrent Th-chemotherapy (18%, versus 13% with Th-corticosteroid combinations and 5% with single-agent Th; p<0.0001 for each comparison). Conclusions– AE information reporting is most complete for events that occur in clinical trials. The FDA should adopt policies that allow for dissemination of adverse events occurring frequently in off-label settings. This policy revision is especially important for Th, as it is the only cancer drug whose use in the recently initiated Medicare demonstration project for oral chemotherapy agents is exclusively off-label, and current FDA regulations prohibit Medicare beneficiaries from receiving AE information describing Th-associated TE. Moreover, the STEPS program currently does not provide patients, pharmacists, or providers with information on TE, and revision of this program should also be considered. Comparison of clinical and diagnostic completeness of TE complications amongst Th-treated cancer patients in the FDA’s AE Reporting System database Clinical Trial Report to Manufacturer, N=49 (%) Non-Clinical Trial Report to Manufacturer, N=131 (%) Report to FDA, N=10 (%) P Administration dates 76% 28% 10% 0.0001 DVT/PE onset-Date(s) 88% 28% 40% 0.0001 Days until DVT/PE 69% 23% 10% 0.0001 Diagnostic-method 59% 17% 40% 0.0001