Adverse drug reactions to the three doses of the severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) mRNA-1273 vaccine in a cohort of cancer patients under active treatment of a tertiary hospital in Madrid, Spain

Study design, eligibility, and study procedures

This observational retrospective study included patients with solid tumours receiving anticancer treatment at the outpatient facility of the Hospital Clínico San Carlos Medical Oncology service in Madrid, Spain. We included all the patients vaccinated with the complete three-dose schedule mRNA-1273 vaccine that were on active anticancer therapy and had complete available information about the date of each vaccination dose and side effects for each of the three doses in electronic medical records. Patient electronic medical records were reviewed retrospectively to collect the data from April 19, 2021 to December 31, 2021.

We selected the patients vaccinated with the complete three-dose schedule mRNA-1273 vaccine from the Preventive Medicine and Public Health Department database. This database was linked with the Medical Oncology Department patient database selecting all oncology patients under active treatment who were in the previous database. From these subjects we selected the ones that had information about the appearance or absence of ADRs after each dose of the vaccine available in electronic medical records. Patients who had a documented SARS-CoV-2 infection or a positive SARS-CoV-2 serology test collected during routine clinical practice in the 7 days prior to the first mRNA-1273 vaccine dose were excluded.

Additional clinical information was abstracted from the electronic medical records, including age, sex, performance status (using Eastern Cooperative Oncology Group (ECOG) performance status score),¹⁷ cancer type, cancer stage and cancer therapy. Authors belonging to Medical Oncology Department had complete access to all electronic medical records available from the patients.

ECOG performance status score describes the level of functioning in terms of the ability to selfcare, daily activity, and physical ability of the patients. ECOG 0 patients are fully active. ECOG 1 patients are not able to perform physically demanding activity but are ambulant and able to perform occupations of light nature. ECOG 2 patients are ambulant, up for more than 50% of waking hours, and capable of all personal care, but are unable to perform any work activities. ECOG 3 patients can perform only limited self-care and are bedridden or confined to a chair for more than half of waking hours. ECOG 4 patients are completely incapacitated, unable to perform any self-care and totally confined to a bed or chair. ECOG 5 patients are dead.¹⁷

Cancer type was divided in seven groups: thoracic malignancies, breast cancer, head and neck cancer, gastrointestinal malignancies, gynaecological malignancies, and others (malignancies that did not belong to any of the previous groups).

Cancer stage was divided in metastatic, patients with malignant lesions in locations organs distant from where the primary tumour is, and the rest, defined as localized.

Cancer therapy was divided in chemotherapy, targeted therapy, immunotherapy and combined therapy (any combination of the previous treatments).

Data analysis

The primary end point of this study was ADRs after each dose of mRNA-1273 vaccine. ADRs were categorized as local adverse reactions which included pain, swelling, rash and itchiness at the site of infection and systemic adverse reactions. Systemic adverse reactions reported by patients were fever (defined as body temperature equal or above 38°C), headache, myalgia, malaise, nausea, arthralgia, chills, adenopathies, urticaria, asthenia and cough.

Data was analysed with IBM SPSS v.25. For descriptive purposes, categorical variables were represented by absolute and relative frequencies and quantitative variables were represented by central and dispersion measures. In order to ascertain the relationship between nominal independent variables (sex, heavily treated status defined as 3 or more lines of previous treatment and past story of systemic adverse drug reactions) we performed a chi-squared test (all expected cell frequencies were greater than five) followed by a Cramér’s V test. For ordinal variables (ECOG performance status score)¹⁷ we performed a Cochran-Armitage test of trend. Significance was tested with an alpha value of 0,05. No multiplicity correction was applied. To test the value as predictors of true baseline variables (age, sex and ECOG performance status) we fitted a binomial logistic regression to ascertain the effects of age, sex and ECOG score on the likelihood that participants have systemic adverse events after the third dose (hereinabove described). We used a Box-Tidwell procedure to evaluate the linear relationship between the logit of the outcome and continuous variables. Following a Bonferroni correction, statistical significance was accepted hen p < 0.0071. Age was included as a continuous variable, sex as a dichotomous variable, being male considered as reference, and ECOG scores were included as categorical variables, with a score of 0 considered the reference.

Ethical considerations

The Comité de Ética del Medicamento e Investigación Clínica (Ethics Committee) of Hospital Clínico San Carlos approved the project with the code: 22/033-E. The Comité de Ética del Medicamento e Investigación Clínica (Ethics Committee) of Hospital Clínico San Carlos deemed the necessary requirements for appropriateness of the protocol in relation to the objectives of the study were met, the informed consent waiver was considered adequate, the procedure foreseen for the handling of personal data was adequate. The ethical precepts formulated in the Declaration of Helsinki of the World Medical Association for medical research on human beings and its subsequent revisions are complied with, as well as those required by the applicable legal regulations according to the characteristics of the study.

Study population

Data was retrieved from electronic medical records on the 31st of December, 2021. In total 93 patients were eligible for the current analysis.¹⁸ Patient demographic, cancer, and therapy characteristics included are summarized in Table 1.

Adverse drug reactions

The number of ADRs categorized as local and systemic after the first, the second and the third vaccine dose are shown in Figure 1. Local ADRs included pain, swelling, rash and itchiness at the site of infection. We can observe that systemic ADRs have a clearly increasing trend while local ADRs have a discrete downward trend.

Figure 1. Number of ADRs after each mRNA-1273 vaccine dose.

After the first, second and third dose 41.9%, 43% and 31.1% of the patients respectively reported a local ADR, while systemic ADRs were increasingly reported after each dose (16.1%, 34.4% and 52.6% of the patients respectively).

The ADRs occurring after the first, the second and the third vaccine dose are described in Figures 2–4 respectively. The most common ADR after the first dose was local reaction with a marked difference compared with systemic ADRs. After the second dose the most frequent ADR was still local reaction, but systemic were becoming increasingly important. After the third dose we can observe that local ADR is less common and takes second place after fever and that the rest of systemic effects are increasingly frequent.

Figure 2. Number and type of adverse drug reactions after 1st dose.

Figure 3. Number and type of adverse drug reactions after 2nd dose.

Figure 4. Number and type of adverse drug reactions after 3rd dose.

No severe ADRs requiring hospitalization occurred in this population. No new SARS-CoV-2 infections occurred in this population during the period studied.

Statistical analysis

There was a statistically significant association between sex and systemic adverse reactions after the third dose, χ²(1) = 12.982, p < 0.001. The association was moderate (Cramer’s V = 0.396). We didn’t find a statistically significant association between heavily treated status and systemic adverse reactions after the third dose, χ²(1) = 0.063, p = 0.802. The association was small (Cramér’s V = 0.05).

Regarding adverse reactions after previous vaccine doses, there was not a statistically significant association between systemic adverse reactions after the first dose and after the third one, χ²(1) = 1.897, p = 0.168, being the association between the variables small, Cramér’s V = 0.172. On the other hand, we found a statistically significant association between systemic adverse reactions after the second dose and systemic adverse reactions after the third dose, χ²(1) = 11.372, p = 0.001. The association was moderate, Cramér’s V = 0.372.

The Cochran-Armitage test of trend showed a statistically significant linear trend, p = 0.012, with a higher ECOG score associated with a lower proportion of patients suffering from systemic side effects. The scores tested were ECOG 0 (n = 56), ECOG 1 (n = 30), ECOG 2 (n = 7), and the proportion of patients suffering a side effect was 0.643, 0.4 and 0.286, respectively.

Concerning the binomial logistic regression model assumptions, age was found to be linearly related to the logit of the dependent variable based on the Box-Tidwell procedure assessment and a Hosmer et al. goodness of fit test was not statistically significant (p = 0.577).¹⁹ The logistic regression model was statistically significant, χ²(4) = 25.641, p < 0.001, explained 32.2% (Nagelkerke R²) of the variance in systemic adverse events after the third dose and correctly classified 72.0% of cases. Sensitivity was 82%, specificity was 60.5%, positive predictive value was 70.7% and negative predictive value was 74.3%. The area under the ROC curve was 0.783 (95% CI, 0.688 to 0.878) (Figure 5), which is at the upper level of an acceptable level of discrimination according to Hosmer et al.¹⁹

Figure 5. Receiving Operating Characteristic curve (ROC curve).

Only two variables were statistically significant: age and sex (as shown in Table 2). Women had 5.79 times higher odds to exhibit systemic adverse events after the third dose compared to males. Increasing age was associated with a decreased likelihood of exhibiting adverse events. ECOG score was not statistically significant but a trend towards a diminished likelihood of systemic adverse reactions with higher ECOG score compared to ECOG 0 was observed.

Table 2. Logistic regression predicting Likelihood of Systemic Adverse Reactions based on Age, Sex and ECOG score.

	B	SE	Wald	df	p	OR	95% CI for OR
							Lower	Upper
ECOG 1	-,434	,538	,652	1	,420	,648	,226	1,858
ECOG 2	-1,669	,967	2,979	1	,084	,188	,028	1,254
Female Sex	1,756	,533	10,839	1	,001	5,789	2,035	16,468
Age (years)	-,059	,024	5,804	1	,016	,943	,899	,989
Constant	2,846	1,524	3,485	1	,062	17,215

Abbreviations: B logarithmic regression slope, SE standard error, degrees of freedom, OR odds ratio, CI confidence interval.

Regarding the variable Sex, male was considered the reference. Regarding the variable ECOG, ECOG 0 was considered the reference.

Underlying data

Dryad: Adverse drug reactions to the three doses of the SARS-COV-2 mRNA-1273 vaccine in a cohort of cancer patients of a tertiary hospital. https://doi.org/10.5061/dryad.cnp5hqc6d.¹⁸

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).