Open access Original research J Immunother Cancer: first published as 10.1136/jitc-2020-001314 on 6 September 2020. Downloaded from A nomogram for predicting mortality in patients with COVID-19 and solid tumors: a multicenter retrospective cohort study Chao Liu,1,2 Li Li,3 Kehan Song,4 Zhi- Ying Zhan,3 Yi Yao,1 Hongyun Gong,1 5 6 7 8 3 Yuan Chen, Qun Wang, Xiaorong Dong, Zhibin Xie, Chun- Quan Ou , Qinyong Hu,1 Qibin Song1 To cite: Liu C, Li L, Song K, ABSTRACT on these factors accurately predicted mortality risk for et al. A nomogram for Background Individualized prediction of mortality risk individual patients. predicting mortality in patients can inform the treatment strategy for patients with with COVID-19 and solid tumors: COVID-19 and solid tumors and potentially improve patient a multicenter retrospective outcomes. We aimed to develop a nomogram for predicting INTRODUCTION cohort study. Journal for COVID-19 poses an unprecedented threat ImmunoTherapy of Cancer in- hospital mortality of patients with COVID-19 with solid 2020;8:e001314. doi:10.1136/ tumors. to global health, with the WHO officially jitc-2020-001314 Methods We enrolled patients with COVID-19 with denoting the outbreak as a ‘public health solid tumors admitted to 32 hospitals in China between emergency of international concern’ on 1 ► Additional material is December 17, 2020, and March 18, 2020. A multivariate March 11, 2020. As of 10 AM CEST on June published online only. To view logistic regression model was constructed via stepwise 19, 2020, 8,385,440 individuals had been please visit the journal online regression analysis, and a nomogram was subsequently confirmed to be infected by SARS- CoV-2, (http:// dx. doi. org/ 10. 1136/ jitc- developed based on the fitted multivariate logistic and 450,686 people had died of COVID-19.2 2020- 001314). regression model. Discrimination and calibration of the nomogram were evaluated by estimating the area under Cancer was newly diagnosed in 18.1 million CL, LL and KS contributed the receiver operator characteristic curve (AUC) for the people and 9.6 million people died of cancer equally. 3 4 model and by bootstrap resampling, a Hosmer- Lemeshow worldwide in 2018. Patients with cancer are test, and visual inspection of the calibration curve. more vulnerable to SARS- CoV-2 infection Accepted 27 August 2020 Results There were 216 patients with COVID-19 with given the relatively high prevalence of under- solid tumors included in the present study, of whom 37 lying chronic disease and cancer- induced and http://jitc.bmj.com/ (17%) died and the other 179 all recovered from COVID-19 anticancer therapy- induced systemic immu- and were discharged. The median age of the enrolled nosuppressive states.5 6 Patients with cancer patients was 63.0 years and 113 (52.3%) were men. who are also infected with SARS- CoV-2 may be Multivariate logistic regression revealed that increasing at higher risk of severe events and deteriorate age (OR=1.08, 95% CI 1.00 to 1.16), receipt of antitumor more rapidly than patients without cancer.7–9 treatment within 3 months before COVID-19 (OR=28.65, 95% CI 3.54 to 231.97), peripheral white blood cell Predicting the prognosis for patients with on September 26, 2021 by guest. Protected copyright. (WBC) count ≥6.93 ×109/L (OR=14.52, 95% CI 2.45 to COVID-19 and cancer can help in prioritizing 86.14), derived neutrophil- to- lymphocyte ratio (dNLR; patient care resources and informing the © Author(s) (or their neutrophil count/(WBC count minus neutrophil count)) choice of treatment strategy, which presum- employer(s)) 2020. Re- use ≥4.19 (OR=18.99, 95% CI 3.58 to 100.65), and dyspnea ably would improve patient outcomes. permitted under CC BY-NC. No on admission (OR=20.38, 95% CI 3.55 to 117.02) were SARS-CoV -2 infection can deregulate commercial re- use. See rights and permissions. Published by associated with elevated mortality risk. The performance immune responses, which may trigger viral BMJ. of the established nomogram was satisfactory, with an hyperinflammation in patients with severe 10 For numbered affiliations see AUC of 0.953 (95% CI 0.908 to 0.997) for the model, COVID-19. Moreover, an uncontrolled end of article. non- significant findings on the Hosmer- Lemeshow test, and overwhelming systemic inflammatory and rough agreement between predicted and observed response can result in acute respiratory distress Correspondence to probabilities as suggested in calibration curves. The syndrome (ARDS), which is the principal Professor Chun- Quan Ou; sensitivity and specificity of the model were 86.4% and ouchunquan@ hotmail. com cause of death among patients with COVID- 92.5%. 11 Conclusion Increasing age, receipt of antitumor treatment 19. Logically, then, indicators of inflamma- Professor Qinyong Hu; tion could be useful for predicting prognosis rm001223@ whu. edu. cn within 3 months before COVID-19 diagnosis, elevated WBC count and dNLR, and having dyspnea on admission in patients with COVID-19. Several important Professor Qibin Song; were independent risk factors for mortality among patients inflammation- related indices derived from qibinsong@ whu. edu. cn with COVID-19 and solid tumors. The nomogram based routine blood tests are generally available for Liu C, et al. J Immunother Cancer 2020;8:e001314. doi:10.1136/jitc-2020-001314 1 Open access J Immunother Cancer: first published as 10.1136/jitc-2020-001314 on 6 September 2020. Downloaded from patients with COVID-19 on admission. Hence, applying Data collection statistical modeling to inflammation-related indices may Information on patient demographics (age and sex), be an economic and effective way to predict disease smoking history (yes/no), cancer characteristics (type, outcomes. stage, receipt of antitumor treatment within 3 months The prognosis for patients with COVID-19 and cancer before COVID-19 diagnosis), comorbid conditions, could also be associated with cancer characteristics signs and symptoms, laboratory and radiology findings, and symptoms on admission. Although mortality is the complications, receipt of steroid, and clinical outcomes endpoint of greatest concern in both cancer and COVID- was extracted from medical records by two physicians 19, little information is available on the possible relevance at each hospital independently. We considered several of cancer characteristics and symptoms on admission in inflammation- related indices derived from routine blood predicting mortality for patients with COVID-19. Only a tests, including peripheral white blood cell (WBC) count, few studies on a small numbers of patients with cancer absolute lymphocyte count, absolute neutrophil count, focused on the outcome of composite severe events neutrophil-to- lymphocyte ratio (NLR), and derived NLR including admission to an intensive care unit, the use (dNLR; neutrophil count/(WBC count minus neutro- of mechanical ventilation, or death, since most patients phil count)). Cancer types included lung, breast, gastro- were still hospitalized at the time of reporting the result intestinal, gynecologic, head and neck, and urogenital. of studies. A previous study of 28 patients indicated that Antitumor treatments included surgery, chemotherapy, patients with COVID-19 and solid cancer who underwent radiotherapy, targeted therapy, immunotherapy, endo- antitumor treatment within 14 days of the COVID-19 diag- crine therapy, and combination therapy. The follow- up nosis were at greater risk of severe events than those who period ended on May 17, 2020, when outcomes for all 12 did not receive treatment during that interval ; also, a patients had been clearly identified. Cause of death for national analysis of 1590 patients with COVID-19, among each patient was determined by both physicians and whom 18 had a history of cancer, revealed that patients oncologists. All discharged patients met the following receiving surgery or chemotherapy within 1 month of criteria: (1) body temperature had returned to normal the COVID-19 diagnosis had a higher risk of severe 9 for more than 3 days; (2) respiratory symptoms had events. Only one study reported a higher risk of death improved notably; (3) lung imaging findings indicated among those receiving chemotherapy within 4 weeks 13 significant alleviation of acute exudative lesions in the before symptom onset. Elevated immunosuppression or lung; and (4) nasal and pharyngeal swab samples were dysfunction could explain the increase in risk of severe negative for the presence of SARS- CoV-2 on two consecu- events among patients with COVID-19 and cancer who tive occasions at least 24 hours apart.15 received antitumor treatment. Likewise, inflammation- related indices may indicate prognosis for patients with Statistical analysis COVID-19 with cancer as well. However, to the best of our The included patients were classified into two groups knowledge, whether inflammation-related indices can be according to clinical outcomes: non-sur vivors and survi- http://jitc.bmj.com/ used to predict mortality risk among SARS- CoV-2- infected patients with cancer has not been reported. vors. Continuous variables were described as medians and Nomograms have been used extensively predicting IQRs, and categorical variables as counts and percentages. prognosis in cancer, mainly because they offer a user- The Mann- Whitney U test was applied to compare contin- friendly graphic presentation of the estimated probabili- uous variables and cancer stage between two groups, and ties of an event, such as death, that is tailored to individual Fisher’s exact test was used to compare categorical vari- 14 patients. Here, we sought to identify prognostic factors ables between groups. on September 26, 2021 by guest. Protected copyright. for patients with COVID-19 and solid tumors and to We developed a nomogram for predicting in- hospital develop a nomogram for predicting mortality risk in a mortality among patients with COVID-19 and solid tumors multicenter retrospective cohort study of 216 patients in four steps: (1) fitting a univariate logistic regression with COVID-19 and solid tumors in China.