Oren, Ohad

Introduction: The impact of beta blocker use on surgical outcomes of cancer patients receiving immunotherapy is unknown. Hypothesis: Beta blocker use is associated with lower post-operative mortality in cancer patients on immunotherapy. Methods: We used an institutional Advanced Cohort Explorer to identify all patients who underwent any surgical procedure less than 90 days after receiving immunotherapy. Data on the procedure performed as well as the post-operative outcome were collected and analyzed per the pre-operative beta blocker use status. Results: A total of 233 patients underwent surgery between 09/2011-06/2019. Mean age was 64.7 years (range: 16-92). The most common cancer diagnoses were lung (48, 44.4%), melanoma (49, 45.4%) and kidney (11, 10.2%). Immunotherapy medications were pembrolizumab ( 127, 54.5%), nivolumab (51, 21.9%), ipilimumab (43, 18.5%), atezolizumab (11, 4.7%) and avelumab (1, 0.43%). Pre-operative beta blocker use was documented in 140 (60.1%) patients. Most common surgical procedures were skin resection (75, 32.1%), wound repair (47, 20.2%), vascular procedures (36, 15.5%), urethral procedures 16 (6.9%) and sentinel lymph node biopsies (10, 4.3%). The 30-day mortality rate was 10.3% (24 deaths). In a multi-variable logistic regression analysis, pre-operative use of beta blockers was associated with a lower risk of death within 30 days from surgery (OR 0.34, CI 0.13-0.87, P=0.024). 30-day mortality rates were higher in patients undergoing vascular (50%, 17 of 34) versus non-vascular (3.5%, 7 of 199) procedures (P<0.0001) in a Chi Square test. Conclusions: Beta blocker use is associated with a lower 30-day mortality rate after surgical procedures in cancer patients treated with immunotherapy. Further investigation is warranted.

7059 Background: Preliminary data suggests that immune-related adverse events (irAEs) are associated with lower all-cause mortality, presumably due to improved anti-tumor responses. Investigations of large cohorts are needed to establish better understanding of that association. Methods: We reviewed the Mayo Clinic database for all patients who received an immune checkpoint inhibitor (ICI). The primary outcome was all-cause mortality. Descriptive and uni-variate analyses were generated. Results: Between March, 2010 and July, 2019, 3,326 patients received an ICI. The most common irAEs were colitis (287, 8.6%), pneumonitis (238, 7.2%) and hepatitis (227, 6.9%). A total of 933 (28.1%) patients developed at least 1 irAE and 176 (5.3%) patients experienced 2 or more irAEs. Survival analysis demonstrated an association between the number of irAEs and all-cause mortality (log-rank, P < 0.0001), a relationship which was maintained for the 3 most common cancer types (lung, melanoma, renal) and for the individual ICI agents. In patients with lung cancer, colitis (P = 0.04) but not pneumonitis (P = 0.83) was associated with improved overall survival. No association between irEA and all-cause mortality was demonstrated in patients with history of stroke (log-rank, P = 0.12), peripheral artery disease (PAD) (log-rank, P = 0.68) and obesity (log-rank, P = 0.18). In an analysis of pre-ICI body-mass index (BMI), an association between irAE and lower overall mortality was shown in patients with BMI < 30 (log-rank, P < 0.001) and not in those with higher BMIs (log-rank, P = 0.09). The presence of stroke, PAD and obesity were associated with higher all-cause mortality in a survival analysis (P < 0.001). The irAE-mortality association was not modulated by the presence hypertension (log-rank, P < 0.0001), diabetes mellitus (log-rank, P < 0.0001), or heart failure (log-rank, P = 0.006). Conclusions: The development of any irAE is associated with higher overall survival. The presence of numerous cardiovascular disease states neutralizes that association, likely a result of competing causes of mortality although interaction with immune or inflammatory pathways is possible. In addition, pneumonitis is not associated with better overall survival in patients with lung cancer presumably due to compromise of already-tenuous respiratory status.

Abstract Background Currently available drug value frameworks integrate data about a medication's benefits and risks. However, the strength of the research data supporting the use of one medication versus another is highly variable. The greater the number and quality of well-conducted trials underlying a certain intervention, the stronger its potential favorable contribution to public health. Drug value should therefore also be a function of the magnitude of clinical evidence gleaned from high-quality clinical trials. Methods We propose an investigation-driven value framework that determines the merit of a drug based on the overall available evidence, including the rigour of relevant trials. The NCCN Hodgkin Lymphoma guidelines were reviewed and data collected about commonly used regimens for Classic HL. Value scores were determined for each drug combination. Results A model was designed incorporating information about the number of RCTs conducted, the total number of participants enrolled, and the generalizability to real-world populations, in addition to the drug's efficacy, significant toxicity and cost (Figure 1). Practice-changing trials supporting each of the currently-endorsed regimens were collected from the HL NCCN guidelines (ABVD, 4; Stanford V 4; Esc BEACOPP 2). The average number of participants per trial was higher for ABVD (1,126) and Esc BEACOPP (1,890) than for Stanford V (267). The generalizability score for the three compounds was 0.5 (ABVD), 0.25 (Stanford V) and -1 (Esc BEACOPP). The mean 5-year overall survival rate was higher in the case of ABVD (96.2%) and Esc BEACOPP (96.1%) compared with Stanford V (92.0%). The toxicity score was 1.24 (ABVD), 3 (Stanford V) and 2.5 (Esc BEACOPP). Using a multi-factorial value scoring system, ABVD, Esc BEACOPP and Stanford V obtained 11.5, 9.5 and 8.5 points, respectively. A scoring system that focused on traditional factors alone (efficacy, toxicity, cost) yielded different scores (ABVD, 7.7; Stanford V 6.7; Esc BEACOPP, 6.5). Conclusion The robustness of clinical trials investigating anti-cancer regimens deserves special consideration when defining drug value. Three high-quality trial characteristics help stratify common regimens in HL into higher/lower value strata using a simple, easy to use algorithm. Disclosures No relevant conflicts of interest to declare.

Abstract Background In order to fulfil their medical and academic mission, clinical trials have to be available to the very populations they are intended to serve. A broad and homogenous geographic distribution is thus paramount to ensure patient fairness as well as to allow optimal generalizability of trial results to real-world populations. We aimed to investigate trends in the geographic distribution of clinical trials in lymphoma performed over the past 2 decades. Methods We systematically searched the Clinicaltrials.gov registry for phase 3 and 4 clinical trials in lymphoma conducted between 01/01/1997 and 12/31/2017. We divided the time period into two intervals and analysed the geographic distribution of trials during these time periods. The Gini coefficient, a statistical measure of inequality ranging from 0 (equal distribution) to 1 (most unequal distribution), was calculated for each period. Results We found 633 clinical trials in lymphoma conducted between 1997 and 2017 globally. 284 and 349 trials were performed in the 1997-2007 and 2008-2017 time periods, respectively. 127 (50%) and 138 (39.5%) of the trials were done in the US. Hodgkin Lymphoma accounted for 55 (19.3%) and 42 (12.0%) studies in the two time periods. CA, TX, IL and NY were consistently in the upper one-fifth of trial-abundant states (each accounting for 3.5-4.5% of all trials). AK, WY, MT and RI were consistently in the lower one-fifth of trial-abundant states (each accounting for 0.13-0.84% of trials). The trial participation of NY and AK mildly increased during the time period while that of TX and RI decreased. In each of the time periods, the 12 most trial-intense states accounted for 41.5% of all investigations. The 12 least trial-intense states were responsible for 8.8% of all the trials. The extent of geographic inequality increased over the years with the Gini coefficient rising from 0.246 to 0.3 between the first and second study period. The geographical pattern of state-specific trial intensity was strongly mitigated after adjusting for each state's population size. Conclusions The geographic distribution of clinical trials in lymphoma in the U.S is not uniform. Our data suggests that the trend of geographic mal-distribution is worsening, a phenomenon that may lead to deleterious public health outcomes. As patients' enrolment continues to be linked to geographic factors, patients' ability to enter clinical trials may be limited and the representation of real-world participants may be skewed (towards particular zip codes, socio-economical status and education levels, etc), consequently affecting the yield and generalizability of clinical trials. Innovative solutions must be devised to minimise any barriers to trial participation for suitable patients. Disclosures No relevant conflicts of interest to declare.

Abstract Background: In addition to a drug's anti-tumor efficacy and tissue toxicity, it is now apparent that affordability is a pivotal factor determining the net therapeutic value of an anti-cancer compound. Cost-Effectiveness Analyses (CEA) are therefore key to appreciating the overall balance between the clinical and economical repercussions of a pharmaceutical agent. We aimed to investigate the cost-effectiveness data of newly approved drugs in cancer care. Methods: The FDA website was reviewed for all drugs approved for any cancer indication between the years 2015-2017. A systematic search of Pubmed and Google Scholar was conducted for Cost-Effectiveness Analyses (CEA) of each of these medications. Average Wholesale Prices were collected from Uptodate.com. Results: 30 drugs were approved for cancer indications in 2015-2017. Each of the approved drugs had an average of 1.27 CEA studies (range 0-6). 4 of the 38 (10.5%) available CEAs were done before approval of an index drug. 18 (60%) of the approved medications had an available published CEA at the time of our search (March, 2018). Partition survival modelling (31.5%) and Markov modelling (26.3%) were the most common methods of analysis. 47.3% (18/38) of the CEAs were thought to reflect a favorable ICER per the analysis's investigators. The nature of the malignancy (solid versus hematologic) did not distinguish between the mean number of a drug's CEAs (1.25; 1.2) but drugs for solid cancer indications were less likely to be associated with a favorable CE assessment compared with hematological cancers (47.8% versus 58.3%). "Expensive" drugs (defined as monthly cost greater than 8,618 dollars) had a lower mean number of CEA studies than "less expensive" drugs (1.09 versus 1.36) and were less likely to be associated with a favorable cost-effectiveness profile (27.2% versus 73.6%). 47.3% (18/38) of the economic evaluations were published as conference proceeding/abstract, 36.8% (14/38) were in the form of a full-text article, and the remainder as journal letters or conference posters (6/38; 15.7%). Conclusion: Insufficient data on the cost-effectiveness profile of novel anti-cancer medications jeopardises our ability to determine their real value. Although these drugs are being routinely used in a large-scale fashion, gaps persist as to their financial harm relative to the associated (potential) clinical benefit. Our study shows that recently-approved anti-cancer drugs have a very small number of CEAs to back their clinical-societal merit. Paradoxically, more costly drugs have fewer CEAs compared with cheaper drugs. Also of note, the majority of the analyses followed the approval of the drug, were published in a non full-text format, and were associated with non-sustainable ICER values. We argue that publication of rigorous, peer-reviewed CEAs should be a mandatory pre-approval step for academia/industry and serve as a pre-requisite to the routine distribution and usage of new cancer-directed medications. Disclosures No relevant conflicts of interest to declare.

Abstract Background The drug value movement aims to assess the clinical-economical merit of anti-cancer medications. A major caveat of quantifying the overall value of an intervention, however, is that patients with distinct clinical features may derive different degrees of benefit: elderly and frail patients with comorbid conditions are more likely to develop early toxicity than younger and otherwise healthier individuals who are more likely to survive for longer periods of time and develop late effects. In addition, patients may prioritize survival to early side effects or to late sequel, or vice versa, which further mandates a more individualised assessment of drug value. We estimated the value of two common chemotherapy combinations used in Hodgkin Lymphoma (HL), as experienced by patients in different age groups, in the early versus the late phases of care. The data were derived from the hematological literature. Methods Pubmed was reviewed for long-term prospective studies of chemotherapy combinations in HL. Data about overall survival rates, incidences of early and late adverse effects, was collected. A scoring system that incorporates efficacy, toxicity and cost was devised (Table 1). Efficacy was derived from overall survival rates at 4 years. Toxicity was categorized as early-serious, early-non-serious, late-serious, late-non-serious, based on the typical time of presentation (early, < 2 years from treatment) and associated mortality (severe > 10% mortality). Cost estimates were derived from the average whole price values available in uptodate.com. The various value determinants (efficacy, early toxicity, late toxicity, cost) were integrated into a grading system (Table 1), factoring each component based on the relative patient-reported preference (per patient age group), as documented in a previous study (1). A value score was calculated for (8 cycles of) ABVD and for (4+4 cycles of) BEACOPP in both the early and the late phase of illness in two age groups. Results ABVD and BEACOPP scored differently based on the patient's age group and the phase of care (Tables 2A, 2B). In patients younger than 55 years of age, ABVD achieved an overall score of 25.3 (out of a maximum of 31.7; 79.8%) while BEACOPP scored 24.5 (77.2%). For patients older than 55 years of age, the total scores for ABVD and BEACOPP were 22.6 (71.3%) and 21.45 (63.4%), respectively. Stratification of each age strata to treatment phase showed that ABVD obtained a score superior to that of BEACOPP in the early phase of care (ABVD 90.5% versus 78.1%, younger than 55; 87.5% versus 77.8%, older than 55). In contrast, BEACOPP achieved a higher score than ABVD in the late phase of care (81.6% versus 71.4%, <55; 77.4% versus 73.4%, >55). The superiority of ABVD to BEACOPP in the early phase persisted in both age groups while the advantage of BEACOPP over ABVD in the late phase diminished significantly in the older patient age group. Conclusions Drug value is not a uniform figure along the treatment continuum of cancer but rather reflects a dynamic equilibrium between benefits and risks that a patient may experience over time. The value of a drug may vary between patients with different clinical profiles. It may also change throughout the course of a patient's illness, thus requiring a more nuanced and longitudinal appraisal. In our work, similar overall value scores for ABVD and BEACOPP were translated into meaningfully different estimates in different disease phases (ABVD superior in early phase, BEACOPP superior in late phase). However, the degree of superiority was tightly associated with patient age group, with BEACOPP's value in the late phase exceeding that of ABVD in younger but not in elderly patients. Our work shows that patient age and treatment phase determine a drug's value in a manner reflective of patients' needs and preferences. The impact of these and other factors on the value of anti-cancer drugs demands further investigation. References: 1. Patient and Physician Preferences for Front-Line Treatment of Advanced Stage Hodgkin Lymphoma in Germany, France and the United Kingdom. Blood 2017 130:4082 Disclosures No relevant conflicts of interest to declare.