Christopher Wallis: Hello, welcome to this UroToday Journal Club. Today we’re discussing a recently published paper looking at analyses of risk, racial disparity, and outcomes among US patients with cancer and COVID-19 infections. I’m Chris Wallis, a Fellow in Urologic Oncology at Vanderbilt, and with me is Zach Klaassen, an Assistant Professor in the Division of Urology at MCG.
This is the citation from this paper recently published in JAMA Oncology. As most people who’ve been awake for the last year will know, COVID-19 spread rapidly through the spring of 2020, rapidly encompassing essentially the entirety of the world, and is now on a resurgence.
We have initial data published in February 2020 from cancer patients in China demonstrating that those with active cancer had higher rates of adverse events, such as invasive ventilation, ICU admission, or death, compared to both those with non-cancer and those who had had previously treated cancer.
However, when we looked at more recent studies from the United States, this relationship did not appear to bear out, and both comparisons of patients with and without cancer demonstrated no difference. Among patients with cancer, major surgical intervention and/or systemic therapy were not associated with disease severity.
However, it’s important to note though, when we look at data from the COVID Surge Collaborative, there are easily identifiable predictors of 30-day mortality among patients with pre or postoperative COVID diagnoses. These include men, older patients, sicker patients, those requiring emergency surgery, and those undergoing cancer surgery. Similarly, when we look at the characteristics of patients admitted to ICUs in Italy, we see that the preponderance of these patients are older, comorbid, and male. These are characteristics that are important for urologic oncologists, given these represent our patients.
The research question in this study was to assess how race and other demographic factors affect the risk of COVID-19 infection as well as outcomes in patients with cancer in the United States.
To do so, the authors performed a retrospective case-control study, and they used the IBM Watson Health Explorys platform, and this allows access to electronic health records from 360 hospitals and over 300,000 clinicians in 50 States in the US encompassing 20% of the US population. Utilizing this approach, the authors identified just over 70 million patients of whom 2.5 million had cancer and 273,000 had a recent cancer diagnosis. They identified just over 16,000 patients with COVID and 1,200 patients who had both a cancer diagnosis and COVID. And finally, 690 who had cancer within the last year as well as a COVID infection. They considered these common cancers, including bladder, breast, colorectal, endometrial, kidney, leukemia, non-Hodgkin’s lymphoma, prostate, liver, lung, melanoma, pancreatic, and thyroid cancer.
They sought to assess essentially three related research questions with individual analyses. The first analysis examined the association between a history of cancer and the risk of developing COVID-19 as stratified by each tumor site, and they adjusted these analyses for the patient age, sex, race, comorbidity, history of cancer treatment, transplant, as well as nursing home exposure.
Secondarily, they assessed the association between demographic factors and the risk of developing COVID among cancer patients alone. Here they considered gender, male versus female, age, looking at younger patients versus seniors, and race, looking at the particular risk for African-American patients.
Finally, they assessed the association between cancer and COVID-19 versus COVID alone versus cancer alone on rates of hospitalization and death. To do so, they used the Cochran-Mantel-Haenszel method for calculating adjusted odds ratios, and they controlled all of their analyses for age, sex, race, co-morbidity, prior transplantation, and nursing homestays. Other potentially relevant demographic factors were excluded because the authors, despite their extremely large sample size, had small samples in these subgroups. Given the multiple analyses performed, they used the Bonferroni correction.
I’ll now pass it over to Dr. Klaassen who’s going to walk us through the results and implications of this study.
Zachary Klaassen: Thanks, Chris. So this first table is the patient characteristics included in this trial. For the discussion of this slide, I’ll focus your attention on the green box, which is the far-right columns. The first column in the left of this box is patients with COVID-19 and all common cancers with a sample size of 1,200. In the far right column is patients with COVID-19 plus a recent, within one year, common cancer with a sample size of 690. You can see here that in both of these groups, females were more commonly affected at more than 50% compared to males around 41 to 43%. The ages were more commonly, in both groups, seniors, more than 65 years of age at 57.5% in the all-cohort and 53.6% in the recent cohort. In terms of race and ethnicity, a majority of these patients were Caucasian at 55%, secondly by African-American at 40%. In terms of insurance status, a majority of these patients did have private insurance followed by Medicare insurance at about 13%.
This figure looks at the association of COVID-19 with recent and all cancer diagnoses. As you can see at the top in figure A, all cancer exposure with the outcome being COVID-19 and the adjusted odds ratio of 1.46 with a significant confidence interval of 1.42 to 1.50. In terms of recent cancer within one year with the outcome being COVID-19, there’s a very strong adjusted odds ratio of 7.14 and 95% confidence interval, 6.91 to 7.39. So certainly here showing that a recent cancer diagnosis does predispose patients to a COVID-19 infection.
In panel B below that, there is an association between different types of cancer and COVID-19 infection. Just to briefly summarize, this is significant for every single one of these cancers. You can see here leukemia, adjusted odds ratio of 12.16, very predisposing to COVID-19. Also, a strong association with non-Hodgkin’s lymphoma with an adjusted odds ratio of 8.54.
This is a similar figure, limited to patients with a recent cancer diagnosis within one year. To summarize once again, all of these cancer exposures are predisposing to a significant adjusted odds ratio, even more so than including all of the malignancies regardless of timing. Here, once again, we see that liver, lung, and leukemia all with adjusted hazard ratios of more than 10, as well as non-Hodgkin’s lymphoma. In terms of urological malignancies, prostate cancer, adjusted odds ratio of 9.64 and bladder cancer, 7.85 and kidney cancer, 7.46. So certainly very large adjusted odds ratios with significant confidence intervals.
This slide here looks at the COVID-19 risk among patients with recent cancer diagnoses but stratified by seniors, African-American versus white, and female versus male. I’ve made it hopefully somewhat intuitive here with the asterisks, as these are the most statistically significant. You can see here that in breast cancer, African-American versus white adjusted odds ratio of 5.44. Looking down at colorectal cancer, African-American versus white, 3.30 adjusted odds ratio. A similar pattern in lung cancer, adjusted odds ratio of 2.53, and prostate cancer, odds ratio of 5.10. And then leukemia, we see that females compared to males had higher adjusted odds ratio of 1.80 and a 95% confidence interval of 1.46 to 2.22.
The final slide of the results looks at hospitalization and death rates among adults. You can see here panel A is hospitalization rates on the left and panel B is death rates on the right. What I’ve also added is the percentages for quick comparison. You can see in the dark blue is African-American and the lighter blue is white. In terms of hospitalization rates for COVID-19 and cancer, 55.6% for African-Americans versus 43.2% versus whites, which was statistically significant. In terms of hospitalization rates for COVID-19 with no cancer, 32.2% for African-Americans versus 19.1% for white. And for hospitalizations of cancer with no COVID-19, 15.7% versus 12.9% for whites. All three of these were statistically significant.
Flipping over to the right panel, which is death rates, you can see that COVID-19 and cancer for African-Americans was 18.5% versus 13.5% of whites, which was not statistically significant. However, both COVID-19 with no cancer and cancer with no COVID-19 both were statistically significant in terms of African-Americans having a higher rate compared to whites.
There are several important discussion points from this observational study, noting that patients with recently diagnosed cancer, specifically leukemia, lung cancer, and non-Hodgkin’s lymphoma, have a significantly increased risk of COVID-19. African-American patients compared to white patients with cancer had a significantly higher risk of COVID-19, with the largest disparity seen in breast cancer with an adjusted odds ratio of 5.44. They also had higher hospitalization rates. However, those with cancer were not more likely to die of COVID-19.
Certainly, we’ve seen in this study that cancer and COVID-19 has a synergistic effect on the patient outcome, with increased hospitalization rates and death rates.
So in conclusion, in this case-control study, patients with cancer were at significantly increased risk for COVID-19 and worse outcomes, which was further exacerbated among the African-American population. However, these associational findings should be replicated and compared to other electronic health medical record databases and registries, such as the COVID-19 and Cancer Consortium Registry. Importantly, these findings highlight the need to protect and monitor patients with cancer as part of a strategy to control the pandemic.
Thank you very much for your attention, and we hope you enjoyed this UroToday Journal Club.