key: cord-309856-flkjl1dm
authors: Westblade, Lars F.; Brar, Gagandeep; Pinheiro, Laura C.; Paidoussis, Demetrios; Rajan, Mangala; Martin, Peter; Goyal, Parag; Sepulveda, Jorge L.; Zhang, Lisa; George, Gary; Liu, Dakai; Whittier, Susan; Plate, Markus; Small, Catherine B.; Rand, Jacob H.; Cushing, Melissa M.; Walsh, Thomas J.; Cooke, Joseph; Safford, Monika M.; Loda, Massimo; Satlin, Michael J.
title: SARS-CoV-2 Viral Load Predicts Mortality in Patients with and Without Cancer Who Are Hospitalized with COVID-19
date: 2020-09-15
journal: Cancer Cell
DOI: 10.1016/j.ccell.2020.09.007
sha: 
doc_id: 309856
cord_uid: flkjl1dm

Patients with cancer may be at increased risk of severe coronavirus disease 2019 (COVID-19), but the role of viral load on this risk is unknown. We measured SARS-CoV-2 viral load using cycle threshold (C T ) values from reverse transcription-polymerase chain reaction assays applied to nasopharyngeal swab specimens in 100 patients with cancer and 2914 without cancer who were admitted to three New York City hospitals. Overall, the in-hospital mortality rate was 38.8% among patients with a high viral load, 24.1% among patients with a medium viral load, and 15.3% among patients with a low viral load (P<0.001). Similar findings were observed in patients with cancer (high, 45.2% mortality; medium, 28.0%; low, 12.1%; P=0.008). Patients with hematologic malignancies had higher median viral loads (C T =25.0) than patients without cancer (C T =29.2; P=0.0039). SARS-CoV-2 viral load results may offer vital prognostic information for patients with and without cancer who are hospitalized with COVID-19.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a devastating received these therapies and 5 days in patients with solid tumors who had not received these 116 therapies. The most common classes of chemotherapy received in patients with hematologic 117 malignancies were antimetabolite drugs (n=6), thalidomide derivatives (n=6), proteasome 118 inhibitors (n=5), and alkylating agents (n=4). An additional seven patients received monoclonal 119 antibodies, five received ibrutinib, three received venetoclax, and two were hematopoietic stem 120 cell transplant recipients.

For data generated using the cobas assay, we used viral load cutoffs based on C T values for the 123 ORF1ab gene target that were previously shown to correlate with in-hospital mortality among 124 hospitalized patients with COVID-19: high, C T value <25; medium, C T value 25-30, low, C T value 125 >30 (Magleby et al., 2020) . For data generated using the Xpert Xpress assay, we established 126 different cutoffs based on C T values for the N2 gene target that are approximately two cycles 127 higher than C T values for the cobas ORF1ab target (Smithgall et al., 2020) : high, C T value <27;

128 medium, C T value 27-32; low, C T value >32. Table 2 ) compared to patients without cancer, even after adjusting for potential confounders. In 136 contrast, having a solid tumor was not associated with having a high viral load in univariate or 137 multivariate analyses compared to patients without cancer. The association between the 138 presence of a hematologic malignancy and a high viral load was also observed in an additional Xpress assays also demonstrated that the presence of a hematologic malignancy, but not solid 142 tumor, was associated with a high admission viral load (aOR 2.90; 95% CI: 1.48-5.69; P=0.002; Table S2 ).

In addition to having a hematologic malignancy, other variables that were independently 146 associated with having a high viral load upon admission (Table 2) 

In the overall cohort, using assay-specific C T value cutoffs, 38.8% of patients with a high viral 156 load died during their hospitalization, compared to 24.1% of patients with a medium viral load, active cancer that adjusted for age and need for supplemental oxygen within 3 hours of 169 presentation to the ED (Table 4) , we found that having a high viral load was independently 170 associated with increased in-hospital mortality (aOR 5.00; 95% CI: 1. 42-8.85; P=0.012) 171 compared to having a low viral load. The risk of in-hospital mortality was also higher in patients 172 with a medium viral load compared to a low viral load, but this association was not statistically 173 significant (aOR 2.13; 95% CI: 0.51-8.85; P=0.30).

In this multicenter observational study of more than 3000 hospitalized patients with COVID-19 in NYC, we found that admission SARS-CoV-2 viral load was highly predictive of in-hospital 178 mortality in patients with and without cancer. This study confirms those of prior reports (Magleby 179 et al., 2020; Pujadas et al., 2020) in a larger cohort and expands upon these prior reports in two 180 important ways. First, both prior studies evaluated the relationship between viral load and 181 mortality using the cobas assay. Others have questioned whether this relationship would persist 182 using different SARS-CoV-2 diagnostic assays and gene targets (Rhoads et al., 2020) . We 183 found that admission viral load was not only associated with mortality using the cobas assay, 184 but was also associated with mortality using the commonly-used Xpert Xpress assay (Table 3) .

The association between viral load and mortality was similar regardless of whether the cobas 186 assay-derived viral load cutoffs were applied to both the cobas and Xpert Xpress assays or if 187 modified viral load cutoffs were applied to the Xpert Xpress assay that incorporate the higher C T 188 values for the Xpert Xpress N2 target compared to the cobas ORF1ab target (Tables 3 and 4, admission viral load was an independent predictor of in-hospital mortality in patients with cancer 194 even after adjusting for important confounders such as age and hypoxia upon arrival to the ED.

Despite the fact that C T values are generated with SARS-CoV-2 RT-PCR diagnostic assays, 196 results of these tests are currently reported as a dichotomous result of detected/positive or not 197 detected/negative. We believe reporting C T values from these assays in patients with and 198 without cancer would provide valuable information that could be used by clinicians to identify 199 patients at high risk of clinical decompensation who may benefit from more intensive monitoring.

This information could also be used when allocating scarce resources, such as the antiviral 201 agent remdesivir (Ison et al, 2020).

Another finding from our study is that patients with hematologic malignancies had higher 204 admission viral loads than patients without cancer; whereas, patients with solid tumors had 205 similar viral loads as patients without cancer. We suspect this finding may be from the 

Additional studies with large sample sizes of patients with hematologic malignancies are needed 231 to more definitely assess whether these patients have increased mortality when hospitalized 232 with COVID-19.

In addition to the presence of a hematologic malignancy, we identified other patient 235 characteristics that were associated with a high admission SARS-CoV-2 viral load, including 236 increased age and comorbidities such as congestive heart failure, diabetes, and chronic kidney 237 disease. We also found that use of inhaled/nasal and oral steroids prior to admission was 238 independently associated with having a high viral load. Although steroid use correlated with 239 chronic lung disease, the association between inhaled or nasal steroid use and high viral load 240 was also observed in a post-hoc analysis limited to patients with asthma or chronic obstructive 241 pulmonary disease. In a randomized trial of dexamethasone in hospitalized patients with clinical significance of this finding is uncertain, particularly given that use of inhaled or nasal 248 steroids was not associated with increased mortality in our study.

A surprising finding was that Hispanic patients were less likely to present with a high viral load, 

A limitation of this study is that we used C T values as surrogate markers for viral load, instead of 257 measuring viral load directly. However, SARS-CoV-2 RT-PCR assays used in clinical 258 laboratories generate C T values, not direct viral load measurements, therefore we believe C T 259 value results have greater potential to be incorporated into patient care. We also only evaluated 260 a single nasopharyngeal swab specimen per patient at the time of hospital admission. Thus, we 261 were unable to assess viral load at the onset of symptoms or changes in viral load over time.

We caution that our findings may not apply to outpatients with COVID-19 who are not sick 263 enough to be hospitalized, as a recent study demonstrated that C T values were similar among 264 symptomatic and asymptomatic SARS-CoV-2-infected patients who did not require 265 hospitalization (Lee et al., 2020c). We encourage subsequent studies to assess the potential 266 role of using SARS-CoV-2 viral load to guide care for outpatients with and without cancer.

Although we have now identified strong associations between admission viral load and mortality 268 using two commonly-used diagnostic platforms, we also encourage investigations of this designed to assess this potential additional role for reporting C T values, but we believe this is an 273 important area of future research. An additional limitation is that we did not capture deaths that 274 occurred after hospital discharge.

In conclusion, using two different diagnostic platforms, we found that admission SARS-CoV-2 277 viral load, as assessed by C T values that are generated by routine RT-PCR diagnostic assays, 278 was highly associated with in-hospital mortality in COVID-19 patients with and without cancer.

Furthermore, patients with hematologic malignancies had higher viral loads than patients 280 without cancer, particularly those who had received chemotherapy or targeted therapies. These 

The Wilcoxon rank-sum test was used for viral load comparisons with 2-sided P values. High viral load is designated as having a C T value <25 using the cobas SARS-CoV-2-specific gene target (ORF1ab) and a C T value 343 <27 using the Xpert Xpress SARS-CoV-2 assay-specific gene target (N2 (Table S1 ). See Table S2 for this analysis using the cobas viral load cutoffs for both the cobas and Xpert Xpress assays. 

See Table S3 for this analysis using the cobas viral load cutoffs for both the cobas and Xpert In a sensitivity analysis that applies the cobas C T value viral load cutoffs (high, C T <25; medium, C T <25-30; low, C T >30) to both the 375 cobas and Xpert Xpress assays, having a high viral load was also independently associated with in-hospital mortality (aOR 4.71; 

We did not use C T value terciles of the N2 gene target from specimens in this study to derive 444 viral load cutoffs for the Xpert Xpress assay because this assay was primarily used after the 445 apex of infections in New York City. Thus, these terciles were not deemed to be representative 446 of the full range of C T values among hospitalized patients throughout the pandemic in NYC.

Data were abstracted manually from the electronic medical record at each hospital using the 450 same quality-controlled protocol at each hospital and entered into a REDCap database (Harris malignancy (e.g., patient with chronic lymphocytic leukemia or prostate cancer). All patients with 460 active cancer who met inclusion criteria were included in the study and we also recorded their

We first compared characteristics and outcomes of patients with and without active cancer who 

COVID-19 mortality

Xpert Xpress SARS-CoV-2 Assay Instructions for Use v1

Viral load dynamics and disease severity in patients infected with SARS-CoV-2 in 605

m1443. REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies: N/A Bacterial and Virus Strains: N/A

Proteins: N/A Critical Commercial Assays cobas SARS-CoV-2 RT-PCR Assay Roche Molecular Systems

Xpert Xpress SARS-CoV-2 RT-PCR Assay Cepheid, Inc. Platform: GeneXpert Infinity Deposited Data: N/A Experimental Models: Cell Lines: N/A Experimental Models: Organisms/Strains: N/A Oligonucleotides: N/A

Recombinant DNA: N/A Software and Algorithms STATA, version