Evidence update – COVID-19 outcomes in cancer patients

Specialty Trainee in Clinical Oncology Dr Ruth Fullerton gives an update on some recent evidence.

For people living with cancer and the clinicians looking after them there remain a lot of uncertainties about the affect of covid-19. There has been particular concern that the anti-cancer treatments we use may put people at increased risk of poorer outcomes should they contract the virus. Throughout the UK big changes have been made in how we manage those with cancer not only in light of this but also taking into account NHS capacity. I am fortunate to work in a centre where we have been able to continue 80-90% of non-surgical cancer treatment throughout the pandemic to date. This has not been the case in other centres and there are clearly many factors at play here.

Ben Goldacre et al from the OpenSAFELY collaborative have conducted a cohort study with the largest sample size to date linking national primary care electronic health records in NHS England with in-hospital COVID-19 death data. The aim was to determine factors associated with risk of death from COVID-19 in the general population.  Sample size is impressive at 17,425,445 and includes all over 18s in NHS England registered with a GP for at least a year previously. This accounts for 40% of the population. There were 5683 deaths attributed to COVID 19.

So what can we learn from this about cancer?  861,192(5%) of the cohort had a non-haematological cancer diagnosis of which 910(0.1%) had an in-hospital covid-19 death.  HR of death was increased in those with a cancer diagnosis within the last 5 years and was most apparent for those diagnosed within the last year.

	CPNS Death HR (95% CI)
Cancer Diagnosis	Age-sex adjusted	Fully adj
Diagnosed <1 year ago	1.83 (1.51-2.21)	1.56 (1.29-1.89)
Diagnosed 1-4.9 years ago	1.39 (1.22-1.58)	1.19 (1.04-1.35)
Diagnosed > 5 years ago	1.03 (0.94 – 1.12)	0.97 (0.88 – 1.06)

Those with a haematological malignancy had a more marked increased risk of in hospital covid-19 death:   >3 fold up to 5 years from diagnosis and nearly double the risk thereafter

This study can also be used to guide how we risk assess our patients based on their co morbidities. Other major risk factors for death included being male, older age, deprivation, uncontrolled diabetes and severe asthma(See figure). As with previous data all non white ethnicity groups had higher risk than those with white ethnicity. This remained true on multivariate analysis taking into account co-morbidity and deprivation.  There was weak evidence for a slightly lower risk of death in smokers.

It is important to note that this looks at in hospital deaths with positive covid-19 test results so will likely be an underestimation. There was a lag in ONS(Office for National Statistics) data which was used for death by other cause of 9 days which wasn’t accounted for. Also the ONS figures will have presumed covid-19 deaths that will not have been included. This highlights the complexities in how we interpret death rates already highlighted by Dr Figueroa.

This study suggests that if you contract covid-19 and have had a cancer diagnosis within the last 5 years you have an increased risk of in-hospital covid-19 death compared to someone without a cancer diagnosis.  A ‘cancer diagnosis’ clearly covers an incredibly wide range of diseases with varying impacts on the individual and potential treatment options so what we need is further evidence within this population.  I think that as time goes on this collaborative has great potential to provide further useful information to help pick this apart and further guide our practice.

A lot of our early evidence of potential risk with systemic cancer treatments comes from studies with small sample sizes. The largest cohort of cancer patients with covid-19 that I have come across is from Memorial Sloan Kettering Cancer Centre (a 514 bed tertiary cancer centre in New York City with approximately 23’000 individuals under active treatment per year). They looked at 423 patients who tested positive for covid19 over a 4 week period excluding asymptomatic patients picked up as part of screening.  168(40%) of patients were hospitalised 84(20%) developed severe respiratory illness and 39(9%) required invasive ventilation.  Overall case fatality rate was 9%.

Of note those with metastatic disease and receipt of chemo within 30days were not associated with increased hospital admission or severe respiratory illness. This conflicts with earlier data in smaller groups of patients.

They did find that use of Immune checkpoint inhibitors(ICIs) within 90 days was associated with both increased hospital admission (HR 3.06 95%CI 1.35-7.2 p=0.007) and severe respiratory illness (HR 3.03 95% CI 1.53-5.98 p=0.001). Again this is not in keeping with evidence within smaller cohorts.  It is important to note that within the cohort only 31 were on these agents. They also concluded that this was true of both lung cancer and non-lung cancer patients as frequency of adverse outcomes were higher in both these groups when the patients were on ICIs.  I don’t feel this was enough evidence to make this conclusion and there may be other confounding factors especially for the lung cancer patients (surgery, pre-existing lung disease, prior radiotherapy, etc).  The suggested explanation for this was exacerbation of ICI related lung injury or ICI triggered immune dis-regulation by T cell hyperactivation that in turn may facilitate covid associated ARDS.

It is clear from these papers that current evidence is conflicting. We had to make fast decisions based on limited evidence with the aim of doing least harm. As we all adapt to a new normality and gain further experience of covid-19 and its impact we need evidence basis to guide our ongoing cancer care.