Letter of response to comments on: Developing a risk assessment score for patients with cancer during COVID-19 pandemic: A war on two fronts.

Bautista et al. presented rational and provoking critiques on our recently published work on developing a score to weight the risk of the novel coronavirus disease 2019 (COVID-19) in patients with cancer [1,2].

The authors arouse questions regarding the methodology of our risk model's development because of the lack of some essential premises (e.g. study design, statistical performance) in the ‘Milano-Policlinico ONCOVID Score’ [1]. Similarly, the authors underline that the selection of candidate predictors in the model should have started from prospective or retrospective data sets, with subsequent validation of the performance of the model itself. These observations are reasonable, and should ideally be respected to warrant adequate methodology of risk assessment. However, COVID-19-related emergency situation has substantially modified several issues in oncology practice, and clinical research has been no exception to this phenomenon. At the time of our score conceptualisation, only few data from small case series of cancer patients with COVID-19 were available. This accounts for the forthcoming validation of the score, which has not been possible until recent times. This rapidly evolving situation has hindered homogeneous prospective data collection. Conversely, it has stimulated researchers to expose relevant findings as soon as they became available, either as a result of literature research or of individual experiences. This was precisely the nature and the aim of our score.

The authors state that a clear definition of the outcome is missing in our work. The aim of our score is to estimate the risk of patients with cancer for severe adverse events or death in case of COVID-19, but also in case of cancer treatment delays (including disease progression, or tumour-related complications). They also enquire on the methodology of our research for variables’ selection. We performed a systematic analysis of the existing evidence, searching PubMed (MEDLINE) from April 1st to 20th, 2020. We sought to identify all English language research studies evaluating COVID-19 and cancer. The literature was queried using the following terms: “SARS-CoV-2” or “COVID-19” AND “cancer” or “oncology” AND “risk assessment” or “risk prediction”. References of included studies were hand-searched to identify potentially relevant adjunctive articles.

Regarding the proposed variables, the authors point out that some clinical (performance status, and corticosteroid use), and laboratory variables were chosen arbitrarily. Our choice was driven by the selection of shared features determining prognosis of patients with COVID-19 and that of cancer patients. Moreover, evidence supports a detrimental role of poor performance status and long-lasting treatment with high-dose corticosteroids in patients with COVID-19 [3,4]. Regarding the selection of laboratory variables, we found appropriate to choose simple and reproducible laboratory parameters which are commonly used in everyday clinical practice in oncology [5]. Moreover, alteration of the selected laboratory parameters demonstrated its clinical significance not only in patients with cancer but also in case of COVID-19, representing a common factor of the two disease entities.

The authors point out the categories of variables and subsequent recommendations were arbitrarily chosen and not solidly sustained. However, the scoring system and the cutoffs were chosen as a result of literature review selection. As specified in our work, given the lack of data in this field at the moment of the score conceptualisation, and the absence of validation in the clinical setting, the assumption of a precise role of these variables should be considered only conceptual. As so, our score should be meant as a proposal for risk categorisation of patients with cancer during COVID-19 emergency, which needs validation before every recommendation became effective. Indeed, the proposed recommendations represent points of reflection for further work, rather than guidelines for risk assessment and patients’ management.

Efforts to validate the design and the methodology of the present score are currently underway. The validation cohort was obtained from one of the largest epidemiological studies to date on the characteristics of cancer patients with COVID-19, reporting data on clinical and demographic predictors of increased mortality in case of infection [6]. As suggested by the authors, the combination of simple clinical features can support oncologists in risk stratification, with the aim to avoid indiscriminate referral of anticancer therapy [6]. Specifically, the targeted population for our score's validation includes adult patients with solid tumours receiving active anticancer treatment in Europe. Validation of the score in the clinical setting will allow us to reconsider the role of the proposed variables, as well as that of the scoring system and of practical recommendations. Preservation of oncological outcomes has been a predominant challenge for oncologists from the beginning of the pandemic. Multidisciplinary cooperation among researchers is a key feature to step forwards into the current coexistence of COVID-19 and cancer, to learn how to fight this war on both fronts.

Conflict of interest statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.