Mortality and Advanced Support Requirement for original reports Patients With Cancer With COVID-19: A Mathematical Dynamic Model for Latin America Alejandro Ruiz-Patiño, MD1,2; Oscar Arrieta, MD3; Luis E. Pino, MD4; Christian Rolfo, MD, PhD, MBA, Drhc5; Luisa Ricaurte, MD1,2; Gonzalo Recondo, MD6; Zyanya-Lucia Zatarain-Barron, MD3; Luis Corrales, MD7; Claudio Martı́n, MD8; Feliciano Barrón, MD3; Carlos Vargas, MD1,2,9; Hernán Carranza, MD1,2,9; Jorge Otero, MD1,2,9; July Rodriguez, MSc1,2; Carolina Sotelo, MD1,2; Lucia Viola, MD10; Alessandro Russo, MD, PhD5,11; Rafael Rosell, MD12,13; and Andrés F. Cardona, MD1,2,9 abstract PURPOSE In the midst of a global pandemic, evidence suggests that similar to other severe respiratory viral infections, patients with cancer are at higher risk of becoming infected by COVID-19 and have a poorer prognosis. METHODS We have modeled the mortality and the intensive care unit (ICU) requirement for the care of patients with cancer infected with COVID-19 in Latin America. A dynamic multistate Markov model was constructed. Transition probabilities were estimated on the basis of published reports for cumulative probability of com- plications. Basic reproductive number (R0) values were modeled with R using the EpiEstim package. Esti- mations of days of ICU requirement and absolute mortality were calculated by imputing number of cumulative cases in the Markov model. RESULTS Estimated median time of ICU requirement was 12.7 days, median time to mortality was 16.3 days after infection, and median time to severe event was 8.1 days. Peak ICU occupancy for patients with cancer was calculated at 16 days after infection. Deterministic sensitivity analysis revealed an interval for mortality between 18.5% and 30.4%. With the actual incidence tendency, Latin America would be expected to lose approximately 111,725 patients with cancer to SARS-CoV-2 (range, 87,116-143,154 patients) by the 60th day since the start of the outbreak. Losses calculated vary between , 1% to 17.6% of all patients with cancer in the region. CONCLUSION Cancer-related cases and deaths attributable to SARS-CoV-2 will put a great strain on health care systems in Latin America. Early implementation of interventions on the basis of data given by disease modeling could mitigate both infections and deaths among patients with cancer. JCO Global Oncol 6:752-759. © 2020 by American Society of Clinical Oncology Licensed under the Creative Commons Attribution 4.0 License INTRODUCTION Three and a half months after the start of the pan- A new pathogen, identified as a novel coronavirus demic, 6,127 cases and 83 deaths have been di- (SARS-CoV-2), triggered a pneumonia (COVID-19) agnosed in Latin America, with a current mortality outbreak in December 2019, starting in Wuhan, of 1.3%.2 China, and spreading quickly to 31 provinces in China The number of COVID-19 cases is still rising. The and . 186 countries worldwide. SARS-CoV-2 is a β- median time from first symptoms to acute respiratory ASSOCIATED coronavirus. It shares a genetic sequence and viral distress syndrome (ARDS) is 8 days (interquartile CONTENT structure with both severe acute respiratory syndrome range, 6-12 days).3 The transition to ARDS occurs in Appendix coronavirus (70% similarity), which caused 349 many severe COVID-19 cases. A possible explanation Data Supplement deaths during 2002-2003 in China, and Middle East for this rapid and serious deterioration is the cytokine Author affiliations respiratory syndrome coronavirus (40% similarity).1 release syndrome, or cytokine storm, an over- and support information (if As of the March 23, 2020, a total of 81,507 confirmed production of immune cells and cytokines that leads to applicable) appear at cases of COVID-19, including 3,274 deaths, were rapid multi-organ system failure and fatal damage to the end of this reported in mainland China, Hong Kong, Macao, and tissues of the lungs, kidney, and heart.4 Recently, Wu article. Taiwan. A rapidly growing number of cases had also et al5 described that the main risk factors related to Accepted on April 23, been reported worldwide. The situation was critical negative outcomes after COVID-19 infection are older 2020 and published at for health systems of Italy, Spain, and Iran, which age, neutrophilia, and organ and coagulation dys- ascopubs.org/journal/ go on May 29, 2020: together have had . 10,200 deaths. Mortality varies function. A recent integrative study that included in- DOI https://doi.org/10. significantly according to country: 0.42% in Germany formation from 46,248 COVID-19 cases confirmed 1200/GO.20.00156 and 9.5% and 6.6% in Italy and Spain, respectively. these findings, highlighting the risk of patients with 752 Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. COVID-19 in Patients With Cancer From Latin America CONTEXT Key Objective SARS-CoV-2 infection represents a substantial clinical issue in patients with cancer. However, the number of cases in Latin America has not reached a critical point, and the impact on deaths of oncologic patients has not been determined or estimated. Knowledge Generated Here, we demonstrate the impact on patients with cancer in terms of both intensive care unit requirements and projected deaths in a variety of scenarios, including a reduction, maintenance, and increase in the R0 value of COVID-19 in Latin America. These findings suggest that the reduction in the reproduction number is a feasible way to reduce cumulative mortality. In addition, with increased cancer disease burden in the region and with incidence and mortality expected to double by the year 2035, control of SARS-CoV-2 in the oncologic population should be a priority. Relevance These results should be used by decision makers and policymakers with the aim of mitigating the negative impact of the COVID-19 pandemic. cardiovascular comorbidity, hypertension, and chronic METHODS lung involvement.6 To estimate the number of ICU requirements per day and Similar to other severe respiratory viral infections, patients per patient as well as expected mortality, a multistate with cancer are more susceptible to COVID-19 infection Markov model was constructed. Disease modeling con- because of their immunosuppression state caused by the sisted of a simulated cohort of patients with cancer infected malignancy and anticancer treatments.7,8 Liang et al9 re- with SARS-CoV-2 that could potentially traverse 4 health ported 18 cases of patients with cancer and COVID-19 states: actively infected without complications, complicated (1% of 1,590 patients); the most frequent neoplasm was infections (ICU requirement), recovered from infection, and lung cancer (28%). Four of 16 patients with cancer and deceased. Mortality attributable to other aspects, such as COVID-19 had received chemotherapy or had undergone disease progression, was not considered in the model surgery in the past month, and the other 12 were survivors because no report of this phenomenon was documented in in routine follow-up after primary resection. Compared with the published cohorts.12,13 Figure 1 represents the model patients without cancer, patients with cancer and COVID- construction as well as the transition paths for patients. The 19 were older (mean age, 63.1 years), were more likely to model was constructed using the heemod package for R have a smoking history (22%), had more polypnea/dyspnea version 3.6.3 (The R Foundation, Vienna, Austria). (47%), and had more severe baseline computed tomog- Transition probabilities were estimated on the basis of raphy scan manifestation (94%). Remarkably, patients with published reports for cumulative probability of complica- cancer were observed to have a higher risk of severe events tions, defined by requiring ICU, mechanical ventilation, or compared with patients without cancer (39% v 8%) even death. Because the behavior of probability of complication after adjusting for other risk factors, including age, smoking changes depending on the number of days after devel- history, and other comorbidities. This suggests that cancer opment of symptoms, a time-varying transition probability history represents the highest risk factor for severe events. was fitted to a distribution obtained by the behavior of an In addition, patients with cancer and COVID-19 de- infected cancer cohort.8 Probability of complication was teriorated more rapidly (median time to severe events, 13 v estimated by extracting the number of patients at risk for 43 days; P , .0001). each event and calculating a hazard function, which was Recently, Kucharski et al10 combined a stochastic trans- later converted to describe transition probability. The dif- mission model with data on cases of COVID-19 in Wuhan ference between the observed and fitted distribution was and international cases that originated in Wuhan to esti- estimated with the Kolmogorov-Smirnov test. The Data mate how transmission had varied over time since January Supplement shows the estimated hazard function and fitted 2020. In addition, Grasselli et al11 published a linear model probability function. Mortality as well as uneventful recovery forecast to estimate critical care utilization for the outbreak probabilities were estimated on the basis of published in Lombardy, Italy. In consideration of the importance of cohorts.12,13 this information and the impact on public health for Latin The number of cancer cases among infected patients was America, we have modeled the mortality and intensive care estimated to be approximately 1.32% (95 CI, 0.98% to unit (ICU) requirement for the care of Hispanic patients with 1.66%).8,12,13 To estimate the full extent and number of cancer infected with COVID-19. possible cancer cases, hospitalizations in the ICU, and JCO Global Oncology 753 Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. Ruiz-Patiño et al C Active infection pmP 0.8 1 Complication Death Uneventful recovery Mortality C 1 Recovery FIG 1. Disease state model. C, complementary probability; pmP, probability for complication. deaths, projections for all Latin American nations, where As calculated by the model, compared with previously data were available, were constructed. Basic reproduction published metrics, median time from diagnosis to death number (R0) values were modeled with R using the Epi- (16.0 days; range, 9.0-22.3 days) and median time from Estim package. Data for parameter estimation was obtained diagnosis to severe event (7.0 days; range, 5.0-15.0 days)13 from the Center for Systems Science and Engineering at are concordant. R0 values were estimated for 15 nations for Johns Hopkins University.14,15 The cutoff date was March which data were available and enough time and cases 27, 2020. Serial interval distribution was modeled after since initial infection had occurred. Estimated incidence a log-normal distribution with a mean value of 4.0 days and R0 variation across days after the first case are rep- (standard deviation, 2.9 days).16 The mean value of the last resented for all included nations in the Data Supplement. 5 days before the construction of this model was defined as Table 1 lists the included nations as well as the estimated R0 for each nation (R0n). After projecting 60 days into the R0 values and number of cases at day 60 of infection, future using a conservative R0 of 1.5, R0n, and 0.5 added depending on different R0 values. If a tendency of actual to R0n, the total number of expected infected patients with incidence continues, and therefore R0, Brazil would be the cancer was calculated per day. Estimation of days of ICU most affected nation with 27,089,291 infected patients in requirement was calculated by imputing the number of 60 days. Contrary to this, Uruguay would be the least af- cumulative cases in the Markov model. Absolute mortality fected with 1,387 cases. If strict social distancing and was derived in the same manner. 75 Data Sharing Complication All data, including implemented R code, are available upon Patients (No.) Recovery request. 50 Death RESULTS 25 By fitting a Weibull distribution (shape, 1.5; scale, 55; Akaike information criterion, 229.0469; P = .43) and protracting the remaining probabilities that aimed to reach an average mortality rate of 25%, behavior dynamics for 0 5 10 15 20 25 30 35 40 patients with cancer infected with SARS-CoV-2 was esti- Days mated. Figure 2 shows a simulated cohort of patients and its transition between disease states. FIG 2. Cohort transition and dynamics among states. 754 © 2020 by American Society of Clinical Oncology Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. COVID-19 in Patients With Cancer From Latin America TABLE 1. Estimated Number of Infected Patients Depending on Varying R0n No. of Cases at 60 Days Nation Estimated R0 (R0n) R0 = 1.5 R0n R0n + 0.5 Argentina 2.67 22,047 3,622,857 23,163,517 Bolivia 1.99 4,066 51,452 670,321 Brazil 3.23 76,398 27,089,291 105,009,995 Chile 1.99 62,423 535,115 4,592,714 Colombia 2.27 38,045 1,178,552 8,762,273 Costa Rica 2.11 10,295 296,800 2,566,206 Cuba 1.98 9,775 146,229 2,624,372 Dominican Republic 2.30 23,264 650,515 4,414,131 Ecuador 2.13 49,293 743,599 5,686,675 Mexico 2.40 12,523 457,142 2,540,246 Panama 2.38 49,256 635,920 4,296,767 Paraguay 2.09 2,660 74,525 1,127,633 Peru 1.87 26,016 188,490 2,069,105 Uruguay 1.03 24,267 1,387 34,135 Venezuela 1.28 10,495 2,368 44,794 Abbreviations: R0, basic reproduction number; R0n, basic reproduction number per nation. active isolation of infected individuals are enforced, with an population, the expected number of cases on the basis of expectation of a reduction of R0 to 1.5, Brazil would remain the previously reported R0 are listed in Table 2. Case the most affected nation with 76,398 cases, but a reduction projection curves are presented for all included nations in of 99.98% of total infections would be achieved. Bolivia the Data Supplement. would benefit the least from a reduction of R0, which would In addition, the expected number of ICU days per person preventing 92.1% of total cases. A third scenario in which computed on the basis of disease dynamics are listed in R0 increases is also presented, being catastrophic for all Appendix Table A1. With taking into account time to death, nations, especially Panama, resulting in the infection of the mortality, and then derived recovery rates and only con- totality of its inhabitants. With regard to the oncologic sidering ICU requirement and not a general hospitalization TABLE 2. Estimated Number of Infected Patients With Cancer by Varying R0n No. of Cases at 60 Days (range) Nation R0 = 1.5 R0n R0n + 0.5 Argentina 291 (216-366) 47,822 (35,504-60,139) 305,758 (227,002-384,514) Bolivia 54 (40-67) 679 (504-854) 8,848 (6,569-11,127) Brazil 1,008 (749-1,268) 357,579 (265,475-449,682) 1,386,132 (1,029,098-1,743,166) Chile 824 (612-1,036) 7,064 (5,244-8,883) 60,624 (45,009-76,239) Colombia 502 (373-632) 15,557 (11,550-19,564) 115,662 (85,870-145,454) Costa Rica 136 (101-171) 3,918 (2,909-4,927) 33,874 (25,149-42,599) Cuba 129 (96-162) 1,930 (1,433-2,427) 34,642 (25,719-43,565) Dominican Republic 307 (228-386) 8,587 (6,375-10,799) 58,267 (43,258-73,275) Ecuador 651 (483-818) 9,816 (7,287-12,344) 75,064 (55,729-94,399) Mexico 165 (123-208) 6,034 (4,480-7,589) 33,531 (24,894-42,168) Panama 650 (483-818) 8,394 (6,232-10,556) 56,717 (42,108-71,326) Paraguay 35 (26-44) 984 (730-1237) 14,885 (11,051-18,719) Peru 343 (255-432) 2,488 (1,847-3,129) 27,312 (20,277-34,347) Uruguay 320 (238-03) 18 (14-23) 451 (335-567) Venezuela 139 (103-174) 31 (23-39) 591 (439-744) Abbreviations: R0, basic reproduction number; R0n, basic reproduction number per nation. JCO Global Oncology 755 Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. Ruiz-Patiño et al stay, the mean number of ICU days per person is ap- scenarios put a loss of , 1% to 17.6% of all patients with proximately 4.1. Estimated mortality, by taking into account cancer in the region (Fig 3). rate sensitivity analysis, is listed in Appendix Table A2. The number of deceased patients varies greatly across the DISCUSSION region. Even in the highest projected R0, nations such as Cancer in Latin America represents an extensive disease Uruguay and Venezuela are expected to have relatively few burden. Projections estimate that by the year 2030, there cases compared with other countries. will be an increase in cases of 72% with 78% mortality By maintaining actual R0n values, Latin America as a re- compared with 2012 data. By that date, it is expected for gion would be expected to lose approximately 111,725 1,831,300 new cancer cases and 1,061,500 cancer- patients with cancer to SARS-CoV-2 (range, 87,116- related deaths to occur.18 With an estimate of 111,725 143,154 patients) by the 60th day since the start of the additional deaths of patients with cancer as a result of outbreak. With an estimated prevalence of cancer in the severe SARS-CoV-2 infection in the next 60 days, an in- included nations of 3,133,806 individuals,17 the various crease of 18.5% in cancer-related deaths, compared with 0-1,000 1,001-2,000 2,001-5,000 > 5,000 FIG 3. Map for Latin American distribution of cases among patients with cancer and expected mortality. COVID- 19 mortality rate for patients with cancer estimated for Latin America. Circle size corresponds to number of cases. Percentage of blue for each nation corresponds to the proportion of all patients with cancer in each nation who could die as a result of SARS-CoV-2 on base 10. (A total blue circle therefore corresponds to 10% of all patients with cancer.) 756 © 2020 by American Society of Clinical Oncology Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. COVID-19 in Patients With Cancer From Latin America 2012 (603,300 deaths), is expected. These sobering disease modeling in real time. First, there are ethical issues predictions further highlight the need for a strict global related to the use of surveillance data that can generate registry of deceased patients with cancer in the setting of data-sharing limitations. Opportunely, data availability on this pandemic to obtain reliable and accurate cancer COVID-19 for modeling and monitoring has not been an statistics that will allow the assessment of health policies issue. On the other hand, model interpretation is also and their impact on cancer control and mortality. a complex process in which misinterpretation can lead to Patients with cancer are indubitably more affected by SARS- erroneous decisions.21 Another delicate aspect to consider CoV-2. In the nationwide analysis of patients with cancer is data quality and adequate reports. Although data are infected in China conducted by Liang et al,9 an estimated available for all examined nations, it is difficult to evaluate prevalence of 1% of COVID-19 cases was identified. By the degree of under-registration samples. Testing strategy comparing this value with the cancer incidence for 2015 in implementation is different in each nation, and accessibility China, a relative over-representation of 300% was observed, is not universal.22 Each Latin American nation has con- which indicates a larger susceptibility to infection. A sig- ducted different plans of action for the detection and nificant proportion of SARS-CoV-2 infections were suspected definition of candidates for the application of tests. Taking to have occurred in the health care setting.3,7 Patients with this into consideration, the number of incident cases could cancer are constantly exposed to health care environments be unreliable, and therefore an absolute estimation is as part of their routine care. This, in combination with an uncertain. To compensate for this aspect, this study offers immunologic misbalance, could help to account for the intervals for which decision makers could influence their larger incidence in the cancer population. conduct. Other difficulties, especially in the estimation of mortality, lie on the different demographics of cancer co- Another key aspect to be considered is the limited re- horts. Patients with cancer tend to be older and to suffer sources available to patients with cancer during these from more comorbidities than the average COVID-19 pa- pandemic times. The expected reorganization of health tient, with an increased risk of complications and death care institutions to face the challenges of a massive influx of beyond the cancer comorbidity itself.9,13 Estimation of the acute and complicated patients with COVID-19 will most effect of variability of cohort composition is difficult with the likely have an impact on the health care delivered to limited information available of infected patients with nonrespiratory patients. For example, the availability and cancer. Furthermore, Latin America has a different de- transport of stem cells for transplantation has faced several mographic of patients with cancer, especially with an challenges. On the one hand, travel restrictions pose dif- infection-related diagnosis, which leads to a younger on- ficulty in the transportation of material and viability of the cologic population.18 To compensate for this variation, cells. On the other hand, donor consecution has been deterministic sensitivity analysis in mortality was in- problematic because donors are required to assist in the corporated, and values within ranges were given. Although screening process as well as spend time in hospitals during ideally, other forms of sensitivity analyses can be con- donation, increasing the risk of exposure to SARS-CoV-2.19 ducted, because of variable uncertainty, these approxi- As more resources are allocated to treating critical patients, mations could be equivocal, especially in resampling it is also understandable that shortages exist in other areas, distributions among demographic characteristics. although this has not been the case up to this moment.19 Anticancer treatments are associated with worse clinical Interesting discrepancies can also be observed. In the case outcomes when administered within 14 days of infection. of Mexico, a nation with an estimated R0 of 2.4, a relatively Zhang et al13 showed a clear impact in severe event–free smaller number of expected cases is observed. Taking into probability compared with patients in whom treatment was account a low initial R0 value as well as a small number of administered after that time frame (P = .037). Taking this cases at day 28, after the projections were estimated, only into consideration, the decision to withhold treatments 32 days of exponential growth at that rate were observed. could be an observable phenomenon, especially in settings Compared with Ecuador, in which 1,403 cases were di- where the risk-benefit of administration is not extensive, agnosed at day 26, results in 34 days of growth with an R0 such as adjuvant treatments in low-risk tumors. Fortu- of 2.13 yielded a much higher case number at day 60. nately, recommendations such as home delivery of treat- Possible explanations for an initial low count of cases could ments, modification of treatment schedules, and specialty include underdiagnosis of early cases. center isolation could be successfully implemented.20 Other aspects to consider include ventilator and ICU bed The utility of modeling disease is extensive and has been availability. To this day, and to our knowledge, no rigorous widely documented. Its main goal is to predict and establish and recent publication discussing the situation on ICU disease behavior. The results from these estimations gain in availability for Latin America exists. Conducts derived from importance in times of pandemics because they offer the information contained in this article should fall to local fundamental data useful for decision making and imple- health authorities. With regard to interventions that could mentation of health policies meant to control, mitigate, and mitigate the impact of this pandemic, social distancing and prevent disease outbreaks. There are several challenges to patient isolation seem to be the most effective ways of JCO Global Oncology 757 Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. Ruiz-Patiño et al diminishing R0 values and reducing the impact on the from around the world have shown that the limits of health number of infections, which protracts this benefit toward care delivery will be pushed, patients with cancer are at patients with cancer.10 a greater risk for complications. On a more positive note, the In conclusion, cancer-related cases and deaths attributable early application of interventions on the basis of data given to SARS-CoV-2 will put a great strain on health care systems by disease modeling could mitigate both infections and in Latin America. Although up to this point experiences deaths among patients with cancer. AFFILIATIONS AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF 1 Foundation for Clinical and Applied Cancer Research, Bogotá, Colombia INTEREST 2 Molecular Oncology and Biology Systems Research Group, Universidad The following represents disclosure information provided by authors of el Bosque, Bogotá, Colombia this manuscript. All relationships are considered compensated unless 3 Thoracic Oncology Unit, Instituto Nacional de Cancerologı́a, Mexico otherwise noted. Relationships are self-held unless noted. I = Immediate City, Mexico Family Member, Inst = My Institution. Relationships may not relate to the 4 Oncology Department, Institute of Oncology, Fundación Santa Fe de subject matter of this manuscript. For more information about ASCO’s Bogotá, Bogotá, Colombia conflict of interest policy, please refer to www.asco.org/rwc or ascopubs. 5 Marlene and Stewart Comprehensive Cancer Center, Experimental org/go/site/misc/authors.html. Therapeutics Program, School of Medicine, University of Maryland, Open Payments is a public database containing information reported by Baltimore, MD companies about payments made to US-licensed physicians (Open 6 Center for Medical Education and Clinical Research, Buenos Aires, Payments). Argentina 7 Department of Oncology, Centro de Investigación y Manejo del Cáncer, Oscar Arrieta San José, Costa Rica Honoraria: AstraZeneca, Merck Sharp & Dohme, Roche 8 Thoracic Oncology Unit, Alexander Fleming Institute, Buenos Aires, Consulting or Advisory Role: Boehringer Ingelheim, Bristol Myers Squibb, Argentina Eli Lilly, Pfizer 9 Clinical and Translational Oncology Group, Clı́nica del Country, Bogotá, Research Funding: AstraZeneca (Inst), Roche (Inst), Merck Sharp & Colombia Dohme (Inst) 10 Thoracic Oncology Unit, Fundación Neurmológica Colombiana, Luis E. Pino Bogotá, Colombia 11 Consulting or Advisory Role: AstraZeneca Medical Oncology Unit, Azienda Ospedaliera Papardo, Messina, Italy 12 Speakers’ Bureau: Roche Coyote Research Group, Pangaea Oncology, Laboratory of Molecular Travel, Accommodations, Expenses: MSD Oncology Biology, Quiron-Dexeus University Institute, Barcelona, Spain 13 Institut d’Investigació en Ciències Germans Trias i Pujol, and Institut Christian Rolfo Català d’Oncologia, Hospital Germans Trias i Pujol, Badalona, Spain Consulting or Advisory Role: Mylan, Archer Biosciences, Oncopass, Inivata, Merck Serono CORRESPONDING AUTHOR Speakers’ Bureau: Novartis, MSD, Boehringer Ingelheim, Guardant Christian Rolfo, MD, PhD, MBA, DrHC, Thoracic Oncology, University of Health, AstraZeneca Maryland Medical Center, Greenebaum Comprehensive Cancer Center, Research Funding: Pfizer 22 South Greene St, Baltimore, MD 21201; Twitter: @ChristianRolfo; Travel, Accommodations, Expenses: AstraZeneca e-mail: christian.rolfo@umm.edu. Uncompensated Relationships: OncoDNA, Biomark, Guardant Health, Gonzalo Recondo EQUAL CONTRIBUTION Consulting or Advisory Role: Roche, Amgen, Pfizer A.R.-P., O.A., L.E.P., C.R., and A.F.C. contributed equally to the study. Travel, Accommodations, Expenses: AstraZeneca, Pfizer Luis Corrales AUTHOR CONTRIBUTIONS Consulting or Advisory Role: AstraZeneca, Roche, MSD Oncology, Pfizer Conception and design: Alejandro Ruiz-Patiño, Oscar Arrieta, Luis E. Pino, Speakers’ Bureau: Roche, AstraZeneca, MSD Oncology Luisa Ricaurte, Zyanya-Lucia Zatarain-Barron, Luis Corrales, Claudio Research Funding: Roche Martı́n, Hernán Carranza, Jorge Otero, July Rodriguez, Andrés F. Cardona Travel, Accommodations, Expenses: Roche, AstraZeneca, MSD Oncology Financial support: Andrés F. Cardona Claudio Martı́n Administrative support: Luis E. Pino, Carlos Vargas, July Rodriguez, Andrés Consulting or Advisory Role: MSD Oncology, Boehringer Ingelheim, F. Cardona AstraZeneca, Bristol Myers Squibb, Pfizer, Roche Provision of study material or patients: Feliciano Barrón, Rafael Rosell, Speakers’ Bureau: Bristol Myers Squibb, Roche, Boehringer Ingelheim, Andrés F. Cardona Pfizer, MSD Oncology, AstraZeneca Collection and assembly of data: Alejandro Ruiz-Patiño, Luis Corrales, Claudio Martı́n, Carlos Vargas, Jorge Otero, Carolina Sotelo, Andrés F. Feliciano Barrón Cardona Speakers’ Bureau: AstraZeneca Data analysis and interpretation: Alejandro Ruiz-Patiño, Oscar Arrieta, Luis Travel, Accommodations, Expenses: AstraZeneca E. Pino, Christian Rolfo, Gonzalo Recondo, Zyanya-Lucia Zatarain- Barron, Claudio Martı́n, Feliciano Barrón, Hernán Carranza, Jorge Otero, Carlos Vargas Lucia Viola, Alessandro Russo, Rafael Rosell, Andrés F. Cardona Honoraria: Roche, Novartis Institutes for BioMedical Research Manuscript writing: All authors Consulting or Advisory Role: Roche, Pfizer, MSD, Bristol Myers Squibb Final approval of manuscript: All authors Speakers’ Bureau: Roche Accountable for all aspects of the work: All authors Travel, Accommodations, Expenses: Roche 758 © 2020 by American Society of Clinical Oncology Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. COVID-19 in Patients With Cancer From Latin America Hernán Carranza Travel, Accommodations, Expenses: Mundipharma, Boehringer Ingelheim, Consulting or Advisory Role: Roche, Amgen, Pfizer AstraZeneca Speakers’ Bureau: Roche Andrés F. Cardona Travel, Accommodations, Expenses: Roche, Merck Honoraria: Bristol Myers Squibb, Roche, Boehringer Ingelheim, AbbVie, Jorge Otero Merck Sharp & Dohme, Celldex Consulting or Advisory Role: AstraZeneca Consulting or Advisory Role: Roche, Merck Sharp & Dohme, Novartis, Speakers’ Bureau: AstraZeneca AstraZeneca, Bristol-Myers Squibb, Foundation Medicine, Boehringer Travel, Accommodations, Expenses: AstraZeneca Ingelheim, Foundation for Clinical and Applied Cancer Research Speakers’ Bureau: Merck Sharp & Dohme, Roche, Bristol Myers Squibb, July Rodriguez Novartis, Foundation for Clinical and Applied Cancer Research, Speakers’ Bureau: Bayer AG, AstraZeneca Foundation Medicine Carolina Sotelo Travel, Accommodations, Expenses: Merck Sharp & Dohme, Bristol Myers Squibb, Roche, Boehringer Ingelheim, Foundation Medicine Travel, Accommodations, Expenses: American Association for Cancer Research No other potential conflicts of interest were reported. Lucia Viola Consulting or Advisory Role: AstraZeneca REFERENCES 1. Xu X, Chen P, Wang J, et al: Evolution of the novel coronavirus from the ongoing Wuhan outbreak and modeling of its spike protein for risk of human transmission. Sci China Life Sci 63:457-460, 2020 2. Center for Systems Science and Engineering, Johns Hopkins University: COVID-19 Dashboard. https://gisanddata.maps.arcgis.com/apps/opsdashboard/ index.html#/bda7594740fd40299423467b48e9ecf6 3. Wang D, Hu B, Hu C, et al. Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan. JAMA 10.1001/ jama.2020.1585 [epub ahead of print on February 7, 2020] 4. Shimabukuro-Vornhagen A, Gödel P, Subklewe M, et al: Cytokine release syndrome. J Immunother Cancer 6:56, 2018 5. Wu C, Chen X, Cai Y, et al: Risk factors associated with acute respiratory distress syndrome and death in patients with coronavirus disease 2019 pneumonia in Wuhan, China. JAMA Intern Med 10.1001/jamainternmed.2020.0994 [epub ahead of print on March 13, 2020] 6. Yang J, Zheng Y, Gou X, et al: Prevalence of comorbidities its effects in coronavirus disease 2019 patients: A systematic review and meta-analysis. Int J Infect Dis 94:91-95, 2020 7. Chemaly RF, Vigil KJ, Saad M, et al: A multicenter study of pandemic influenza A (H1N1) infection in patients with solid tumors in 3 countries: Early therapy improves outcomes. Cancer 118:4627-4633, 2012 8. El Ramahi R, Freifeld A: Epidemiology, diagnosis, treatment, and prevention of influenza infection in oncology patients. J Oncol Pract 15:177-184, 2019 9. Liang W, Guan W, Chen R, et al: Cancer patients in SARS-CoV-2 infection: A nationwide analysis in China. Lancet Oncol 21:335-337, 2020 10. Kucharski AJ, Russell TW, Diamond C, et al: Early dynamics of transmission and control of COVID-19: A mathematical modelling study. Lancet Infect Dis 20:553-558, 2020 11. Grasselli G, Pesenti A, Cecconi M: Critical care utilization for the COVID-19 outbreak in Lombardy, Italy: Early experience and forecast during an emergency response. JAMA 323:1545, 2020 12. Yu J, Ouyang W, Chua MLK: SARS-CoV-2 transmission in patients with cancer at a tertiary care hospital in Wuhan, China. JAMA Oncol 10.1001/jamaoncol. 2020.0980 [epub ahead of print on March 25, 2020] 13. Zhang L, Zhu F, Xie L, et al: Clinical characteristics of COVID-19-infected cancer patients: A retrospective case study in three hospitals within Wuhan, China. Ann Oncol 10.1016/j.annonc.2020.03.296 [epub ahead of print on March 26, 2020] 14. Dong E, Du H, Gardner L: An interactive web-based dashboard to track COVID-19 in real time. Lancet Infect Dis 20:533-534, 2020 15. Nishiura H, Linton NM, Akhmetzhanov AR: Serial interval of novel coronavirus (COVID-19) infections. Int J Infect Dis 93:284-286, 2020 16. The Lancet: GLOBOCAN 2018: Counting the toll of cancer. Lancet 392:985, 2018 17. Curado MP, de Souza DLB: Cancer burden in Latin America and the Caribbean. Ann Glob Health 80:370-377, 2014 18. Burki TK: Cancer care in the time of COVID-19. Lancet Oncol 21:628, 2020 19. You B, Ravaud A, Canivet A, et al: The official French guidelines to protect patients with cancer against SARS-CoV-2 infection. Lancet Oncol 21:619-621, 2020 20. Rivers C, Chretien J-P, Riley S, et al: Using “outbreak science” to strengthen the use of models during epidemics. Nat Commun 10:3102, 2019 21. Salathe M, Althaus CL, Neher R, et al: COVID-19 epidemic in Switzerland: on the importance of testing, contact tracing and isolation. Swiss Med Wkly 150:w20225, 2020. doi: 10.4414/smw.2020.20225 n n n JCO Global Oncology 759 Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms. Ruiz-Patiño et al APPENDIX TABLE A1. Estimated Number of Cumulative ICU Days Per Person for All Affected Patients With Cancer by Varying R0 Values No. of Cumulative ICU Days Per Person at 60 Days Nation Estimated R0 (R0n) R0 = 1.5 R0n R0n + 0.5 Argentina 2.67 1,191 195,838 1,252,153 Bolivia 1.99 220 2,779 36,211 Brazil 3.23 4,127 1,463,389 5,672,740 Chile 1.99 3,372 28,907 248,103 Colombia 2.27 2,055 63,667 473,346 Costa Rica 2.11 556 16,033 138,629 Cuba 1.98 528 7,899 141,771 Dominican Republic 2.3 1,257 35,141 238,456 Ecuador 2.13 2,663 40,170 307,200 Mexico 2.4 677 24,695 137,227 Panama 2.38 2,661 34,353 232,115 Paraguay 2.09 144 4,026 60,916 Peru 1.87 1,405 10,182 111,775 Uruguay 1.03 1,311 75 1,844 Venezuela 1.28 567 128 2,420 Abbreviations: ICU, intensive care unit; R0, basic reproduction number; R0n, basic reproduction number per nation. TABLE A2. Estimated Number of Deceased Patients With Cancer by Varying R0 Values No. of Deceased Patients at 60 Days (range) Nation R0 = 1.5 R0n R0n + 0.5 Argentina 73 (54-88) 11,955 (8,847-14,538) 76,440 (56,565-92,951) Bolivia 13 (10-16) 170 (126-206) 2,212 (1,637-2,690) Brazil 252 (187-307) 89,395 (66,152-108,704) 346,533 (256,4364-421,384) Chile 206 (152-250) 1,766 (1,307-2,147) 15,156 (11,215-18,430) Colombia 126 (93-153) 3,889 (2,878-4,729) 28,916 (21,397-35,161) Costa Rica 34 (25-41) 979 (725-1,191) 8,468 (6,267-10,298) Cuba 32 (24-39) 483 (357-587) 8,660 (6,409-10,531) Dominican Republic 77 (57-93) 2,147 (1,589-2,610) 14,567 (10,779-17,713) Ecuador 163 (120-198) 2,454 (1,816-2,984) 18,766 (13,887-22,819) Mexico 41 (31-50) 1,509 (1,116-1,834) 8,383 (6,203-10,193) Panama 163 (120-198) 2,099 (1,553-2,552) 14,179 (10,493-17,242) Paraguay 9 (6-11) 246 (182-299) 3,721 (2,754-4,525) Peru 86 (64-104) 622 (460-756) 6,828 (5,053-8,303) Uruguay 80 (59-97) 5 (3-6) 113 (83-137) Venezuela 35 (26-42) 8 (6-10) 148 (109-180) Abbreviations: R0, basic reproduction number; R0n, basic reproduction number per nation. 760 © 2020 by American Society of Clinical Oncology Downloaded from ascopubs.org by 77.231.195.252 on July 2, 2020 from 077.231.195.252 Copyright © 2020 American Society of Clinical Oncology. See https://ascopubs.org/go/authors/open-access for reuse terms.