Lack of detail in population-level data impedes analysis of SARS-CoV-2 variants of concern and clinical outcomes

The SARS-CoV-2 lineage B.1.1.7 is characterised by a suite of defining mutations in the immunodominant spike protein, including a signature Asp501Tyr substitution in the receptor-binding domain.¹ First reported in December 2020, in the UK, the variant's discovery coincided with a substantial surge in case numbers and fatalities in the UK, raising concerns that this variant was both more infectious and virulent than previous variants. Epidemiological and modelling studies have yielded good evidence that B.1.1.7 is more transmissible than other variants.1, 2 However, conclusions as to the effects of B.1.1.7 on disease severity are less certain. Confounding factors including health-care resource use, demographic changes, and socio-behavioural trends affect clinical outcomes, including mortality, and are difficult to adjust for without detailed, robust, patient-level data.

In The Lancet Infectious Diseases, Dan Frampton and colleagues³ report their findings from such a study. Analysing a cohort of 341 patients, including 198 (58%) with B.1.1.7 infections, the authors correlated outcomes with granular clinical data. Their observation that B.1.1.7 infections were associated with increased viral loads corroborates findings from two other studies4, 5 and provides a mechanistic hypothesis that increased transmissibility is via increased respiratory shedding. Yet, disease severity and clinical outcomes between patients with B.1.1.7 and non-B.1.1.7 infections were similar after adjusting for differences in age, sex, ethnicity, and comorbidities. Importantly, this study was done from Nov 9, to Dec 20, 2020, before the late-December peak in UK COVID-19 infections, avoiding any confounding effect of the availability of health-care resources on mortality.

This finding is in contrast with three studies that reported increased mortality associated with lineage B.1.1.7 (table ).6, 7, 8 Several factors might explain this discordance. Two of these studies were based on a community-based testing dataset, whereas Frampton and colleagues studied a cohort of patients admitted to hospital, which included substantially more older adults than the other studies did. Although the proportion of patients with severe illness was not reported by the other studies, this proportion was probably much lower than that in Frampton and colleagues' study. Hence, although these large community studies found a significant difference in mortality at a population level, the absolute risk increase affecting individual patients is probably minimal.

Table.

Comparison of studies assessing the effect of lineage B.1.1.7 on disease severity and clinical outcomes

	Frampton et al3	Challen et al6	Davies et al7	Grint et al8
Patient recruitment	Hospitalised patients with confirmed COVID-19	Public health data from community-based testing dataset	Public health data from community-based testing dataset	Public health data from both community and hospital-based testing dataset
Study dates	Nov 9, to Dec 20, 2020	Oct 1, 2020, to Jan 28, 2021	Nov 1, 2020, to Jan 23, 2021	Nov 16, 2020, to Jan 11, 2021
Number of participants	341 (69%) included of 496 available patients screened	109 812 (11·6%) included of 941 518 available patients screened	1 146 534 (51·1%) included of 2 245 263 available patients screened	184 786 (41·9%) included of 441 161 available patients screened
Age of participants, years	Median 60 (IQR 47–75)	Mean 46·3 (SD 11·0)	1–34 (513 726/1 14 6534 [44·8%]); 35–54 (403 313/1 14 6534 [35·2%]); 55–69 (175 983/1 14 6534 [15·3%]); 70–84 (440 46/1 14 6534 [3·8%]); ≥85 (9446/1 14 6534 [0·8%])	Median 38·0 (IQR 24·0–52·0); mean 38·2 (SD 18·1)
Detection of lineage B.1.1.7	Whole-genome sequencing and matching to COG-UK Mutation Explorer database	Surrogate measure using S-gene negativity on Thermo TaqPath COVID-19 multiplex PCR assay	Surrogate measure using S-gene negativity on Thermo TaqPath COVID-19 multiplex PCR assay	Surrogate measure using S-gene negativity on Thermo TaqPath COVID-19 multiplex PCR assay
Controls	Non-B.1.1.7 infections	S-gene positive patients	S-gene positive patients	S-gene positive patients
Primary outcome	Clinical severity as defined by WHO ordinal scale ≥6; mortality at 28 days	Mortality at 28 days	Mortality at 28 days	Mortality at 28 days
Overall rate of severe disease	36·9%	Data not available	Data not available	Data not available
Overall mortality rate	16·2%	0·3%	0·9%	0·5%
Effect on mortality	No significant difference	HR 1·64 (95% CI 1·32–2·04)	HR 1·55 (95% CI 1·39–1·72)	HR 1·67 (95% CI 1·34–2·09)

HR=hazard ratio.

Furthermore, instead of whole-genome sequencing as used by Frampton and colleagues, these studies used S-gene target failure (SGTF) on PCR assay as a surrogate measure for detection of lineage B.1.1.7. The B.1.1.7 variant is associated with nucleotide deletions that prevent S gene target amplification by several commercial tests. This method gave rise to notable selection bias because S-gene data were unavailable in certain areas because of assay availability, with half of patients in both studies having unknown S gene status. Crucially, missing SGTF status was associated with older age and place of residence, with most patients in care homes having absent S-gene data, where mortality rates from COVID-19 are highest.

Thus, although limited by a much smaller dataset, the study by Frampton and colleagues has important advantages over the three community studies. These advantages include the use of whole-genome sequencing, recruitment of hospitalised patients, and a population reflective of the spectrum of severity in whom increased virulence will have the greatest effect on outcomes. The finding that lineage B.1.1.7 infection did not confer increased risk of severe disease and mortality in this high-risk cohort is reassuring but requires further confirmation in larger studies.

These differences between B.1.1.7 and non-B.1.1.7 lineages mirror those of other virological sub-groups of SARS-CoV-2. Similarly conflicting data was initially reported when variants carrying the Asp614Gly substitution emerged and became the dominant variants worldwide over the first few months of the pandemic in 2020. Early population-level data suggested that this substitution was associated with increased COVID-19 mortality but later cohort studies found no effect on disease severity.9, 10 In a study we did in Singapore comparing different SARS-CoV-2 clades, Asp614Gly was associated with increased viral loads without changes in severity or transmission.¹¹

Genetic drift and selection pressures (in particular with passive antibody treatments and vaccination) will continue to engender changes in SARS-CoV-2 and might result in the emergence of variants of high consequence—variants that are more virulent, escape from host immunity, or are resistant to treatment. Active, timely, and broad-based genomic surveillance is crucial for their early detection. But careful epidemiologic and clinical assessment, coupled with a healthy scepticism, is important when assessing claims of the effect of these variants.

BEY reports personal fees from Roche and Sanofi, outside the submitted work. All other authors declare no competing interests.