Clinical course of Coronavirus Disease-19 in patients with haematological malignancies is characterized by a longer time to respiratory deterioration compared to non-haematological ones: results from a case–control study

QuerySevere acute respiratory syndrome coronavirus-2 (SARS-CoV2) was first reported in late December 2019 and was recognized as the causative agent of an acute respiratory tract infection named Coronavirus Disease 2019 (COVID-19) [1]. Although most patients were asymptomatic or exhibited mild respiratory symptoms, approximately 8% may require hospitalization and 1% admission in Intensive Care Unit (ICU) [2]. In this context, patients suffering from haematological malignancies were considered an extremely vulnerable group, with a high risk of severe illness [3, 4] due to the prolonged and severe immunosuppression resulting from both chemo-immunotherapy and the underlying disease, which may weaken the immune system [5]. Indeed, the severe immunosuppression and the potential delay in the underlying disease’s treatment due to SARS-CoV2 diagnosis may affect the outcome of patients with haematological malignancies and COVID-19. On the other hand, these patients might also be protected from severe forms of COVID-19 thanks to the lower immune reaction during SARS-CoV2 infection [6, 7].

Although the relation between haematological malignancies and severity of SARS-CoV2 disease has been widely demonstrated, still no clear data on the risk factors of mortality in this special population exist. Furthermore, it is not fully elucidated whether there are differences in the clinical course of COVID-19 disease between patients with or without haematological malignancies.

As a matter of fact, so far the majority of studies have included mixed cohorts combining patients with haematological and solid cancer [3, 8‐10]. Whereas, case–control studies, specifically matching subjects with and without haematological malignancies affected by SARS-CoV2, are still rare and include only a few patients [11, 12].

Based on these premises, the primary aim of this case–control study was to evaluate the clinical and prognostic features of patients with haematological malignancies and COVID-19. The secondary aim was to analyse risk factors for in-hospital mortality in this special population, with a particular emphasis on the respiratory worsening during hospitalization.

From April 2020 to April 2021, a retrospective, single-center, case–control (1:3) study was performed on patients with COVID-19 hospitalized at Azienda Ospedaliero-Universitaria Policlinico Umberto I, Sapienza University of Rome. Cases were patients with haematological malignancies and COVID-19. Controls were patients with COVID-19 but without haematological malignancies. Cases and controls were matched for sex, age (decades) and time of hospitalization (Supplementary Fig. 1).

Inclusion criteria were: i) diagnosis of COVID-19 by means of molecular analyses, ii) hospitalization, and iii) age > 18 years. Patients with haematological disorders other than malignancy and age < 18 years were excluded from the study.

Nasopharyngeal swab samples were collected and SARS-CoV-2 RNA was detected using real-time RT-PCR assay (RealStar SARS-CoV2 RT-PCR, Altona Diagnostics).

For each subject, laboratory and clinical data at hospital admission, as well as treatment data and SARS-CoV-2 RNA time of detection, were collected and recorded anonymously in an electronic database. As for treatment data, antiviral therapy and the choice of using steroids were based on the national, regional and local guidelines available at the time [13, 14] and on clinical judgement, also following consultation with haematologists when appropriate.

In patients with haematological malignancies, the following variables were also collected: type and status of underlying disease, rate and type of active treatment, prior (30 d) chemotherapy, percentage of total lung parenchyma involvement during the disease by means of CT scan, CT score.

A total of 264 patients were included in the study (66 cases and 198 controls) (Supplementary Fig. 1). General features of the study population are shown in Table 1. Age, gender distribution and period of hospitalization did not differ between cases and controls, reflecting the correctness of the match. Likewise, infection severity and the PaO2/FiO2 at hospital admission were similar between cases and controls [(asymptomatic 4% vs 6%, p = 0.438, moderate infection 64% vs 64%, p = 0.904, severe infection 27% vs 29%, p = 0.796, critical illness 5% vs 6%, p = 0.619, respectively), 357.0 ± 102.1 vs 352.4 ± 101.0, p = 0.750]. Compared to controls, cases had a higher rate of chronic corticosteroids use and infections in the previous 30 d (39% vs 10%, p < 0.001 and 21% vs 3%, p = 0.001, respectively), as well as higher percentage of thrombocytopenia (39% vs 18%, p = 0.002), neutropenia (14% vs 1%, p = 0.003) and lower albumin levels (35.6 ± 6.7 vs 37.4 ± 5.5 g/dL, p = 0.003). C-reactive protein values did not differ between cases and controls (6.57 ± 8.41 vs 7.06 ± 7.52 mg/dL, p = 0.65).

Amongst cases, non-Hodgkin lymphoma (NHL) was observed in 27 (41%) of the subjects, followed by multiple myeloma (9, 14%), acute myeloid leukemia (7, 11%), chronic lymphocytic leukemia (7, 11%) and acute lymphocytic leukemia (5, 8%). Fifteen subjects (23%) have received a new diagnosis of haematological malignancy in the last months and, therefore, have not yet started or concluded treatment for the underlying disease, 34 (52%) were in complete or partial remission, 12 (18%) were suffering a relapsing or refractory disease, while 4 (6.1%) still had a not defined disease status. Prior (90 d) active treatment was given in 36 subjects (54.5%) [chemotherapy, n = 17, immunotherapy, n = 9, chemotherapy + immunotherapy, n = 10] and, amongst them, 12 (18%) had received recent (30 d) chemotherapy. Only four patients (6%) had a history of allogeneic stem cell transplantation (Supplementary Table 1).

Concerning the treatment of SARS-CoV2 infection, no differences emerged regarding the use of remdesivir and enoxaparin between cases and controls (47% vs 40%, p = 0.250 and 73% vs 81%, p = 0.234, respectively). In contrast, corticosteroid and antimicrobial therapies were significantly more frequent in cases than in controls (91% vs 73%, p < 0.001 and 83% vs 65%, p < 0.001, respectively). Of note, also the duration of corticosteroids treatment was longer in patients with haematological malignancies (18.1 ± 22.9 vs 8.1 ± 4.3 days, p < 0.001).

Super-infection rate and colonization by multi-drug resistant pathogens were similar in the two groups, whereas, as expected, opportunistic infections more frequently affected the haematological ones (p = 0.004). Furthermore, cases had a significantly longer duration of hospitalization (31.79 ± 28.04 vs 19.16 ± 13.39 days, p = 0.001) as well as a longer duration of viral shedding (35.4 ± 23.4 vs 19.4 ± 10.2 days, p = 0.010) than controls. Compared to controls, cases experienced higher rates of respiratory deterioration (78.7% vs 65.5%, p = 0.04). Furthermore, the time from symptoms onset, hospital admission and SARS-CoV2 diagnosis to respiratory worsening was significantly longer in cases [13.20 ± 11.69 vs 9.45 ± 5.38 days (p = 0.041), 12.3 ± 12.3 vs 3.3 ± 3.1 days (p = 0.001) and 11.63 ± 13.69 vs 5.69 ± 4.39 days (p = 0.006), respectively], leading to a delayed respiratory deterioration in cases compared to controls. As shown in Fig. 1, 29% of cases developed respiratory worsening more than 10 days after hospital admission, compared to 5% in the control group (Fig. 1, Panel A); likewise, 37% of cases worsened more than 10 days after SARS-CoV2 diagnosis, compared to 11% of controls (Fig. 1, Panel B). Cases had higher rates of ICU admission and in-hospital mortality as compared to controls: 27% vs 8%, (p = 0.002) and 35% vs 10%, (p < 0.001), respectively (Fig. 2).

At multivariable analysis, having an haematological malignancy (i.e. belonging to the case group) [OR 4.76 (2.4–9.4), p < 0.001], thrombocytopenia at admission [OR 3.03 (1.6–5.6), p < 0.001], neutropenia at admission [OR 31.1 (3.86–250.7), p = 0.001], reduced level of serum albumin [OR 3.1 (2.1–6.6), p = 0.001], chronic corticosteroid use [OR 3.65 (1.49–8.96), p = 0.004], prior infection [OR 57.7 (3.2–1014.4), p = 0.006], low PaO2/FiO2 ratio at the time of respiratory worsening [OR 2.2 (1.3–10.4), p = 0.022] and a longer time to respiratory deterioration [OR 3.3 (2–12.4), p = 0.002] were independent risk factors for in-hospital mortality (Table 2).

As shown in Table 3, independent risk factors for mortality in patients with haematological malignancies were neutropenia at hospital admission [OR 3.1 (2.9–270), p < 0.001], prior infection [OR 7.7 (3.2–112), p < 0.001], a longer time to respiratory deterioration [OR 4.1 (2.2–12.4), p < 0.001], active treatment [OR 4.4 (1.5–22.3), p < 0.001], multiple myeloma [OR 1.5 (1.1–3.3), p = 0.044], the variation of the total lung parenchyma involvement (CT1-CT2) [OR 2.6 (1.4–4.8), p = 0.004] and the variation of the CT lung score during hospitalization [OR 2.6 (1.2–12.2), p = 0.006].

In patients under active treatment, those receiving only immunotherapy (n = 9) had a mortality rate of 11.2% (1/9), compared to 29.4% (5/17) and 30% (3/10) of chemotherapy only and chemotherapy + immunotherapy, respectively.

Active treatment [OR 3.2 (1.8–12.3), p < 0.001], the variation of the CT lung score during hospitalization [OR 1.4 (1.1–8.2), p < 0.001], the variation of the total lung parenchyma involvement (CT1-CT2) [OR 1.6 (1.2–6.8), p < 0.001] and non-Hodgkin lymphoma [OR 1.4 (1.1–10.3), p = 0.011] were associated with in-hospital development of respiratory deterioration in cases (Supplementary Table2).

Our data support previous observations that, amongst patients with COVID-19, those with haematological malignancies have higher mortality and ICU admission rates than subjects without malignancies [3, 11, 21‐24]. Furthermore, they have increased risk in the presence of neutropenia at hospital admission, prior infection, active chemotherapy treatment and multiple myeloma [4, 8].

Nevertheless, in our opinion, the most interesting observation of this report was that i) haematological patients experienced a higher rate of respiratory deterioration than controls, ii) that this complication occurred later during the disease and, notably, iii) that the longer time to respiratory deterioration together with the increase of the CT lung score during hospitalization represented independent risk factors for mortality.

To point out the different clinical course between the two groups, we analysed the PaO2/FiO2 ratio at the time of worsening and we found that haematological patients experienced a higher rate of worsening than controls, although the PaO2/FiO2 ratio at the time of deterioration was still higher than in subjects without malignancies, apparently in contrast with the worse observed outcome. Furthermore, deterioration occurred more than 10 days after hospital admission in 29% of cases and only in 5% of controls. Being a novel aspect of the infection’s clinical course, the finding that haematological patients experienced a delayed respiratory worsening still has no clear explanation. Indeed, on one hand the immunocompromised status may lead the virus to replicate and to persist for a longer time; on the other hand, the immune response to the viral infection is delayed [12, 22, 25]. It may also be speculated that, as far as the restoration of the immune system occurs, for instance after chemotherapy-induced amelioration of the haematological disease, patients experience an excessive inflammatory response, which, eventually, is responsible for the clinical observed deterioration.

Another interesting finding of the present study was the correlation between the respiratory worsening and the radiology pattern in patients with haematological malignancies [26]. Although CT severity score has been shown to be a predictor of death in patients with COVID-19 pneumonia [27, 28], very few data are present regarding the role of this radiological parameter in patients with haematological malignancies. In this special context, we showed that not the CT score itself but its change during hospitalization was predictive of worse outcomes in haematological patients and, therefore, it may be used as a feature of clinical worsening.

Corticosteroids use and duration of treatment were significantly higher in cases than in controls, suggesting that the immune system of this special population is not only affected by the malignancy itself but also by its treatment. Likewise, cases had a higher rate of hypoalbuminemia at hospital admission than controls, confirming its prognostic role for mortality, prolonged viral shedding and coagulopathy during COVID-19 pandemic [18, 29‐31] as well as for morbidity and mortality in several malignancies, including the haematological ones [32, 33].

Patients with haematological malignancies often experience prolonged shedding of viral RNA from the upper respiratory tract. In line with this, we showed that duration of viral positivity was significantly longer in patients with haematological malignancies than those without. With the continuous emergence of new variants, this phenomenon may be of significant concern, since it seems to contribute to the selection of sequence SARS-CoV-2 variants within the same host [34‐36].

Our study has several limitations. First, its retrospective and single-centre nature. Second, we could not establish with certainty whether the diagnosis of SARS-CoV2 infection would delay the treatment of the haematological malignancy. This leads to a possible additional reason explaining the observed high mortality. Furthermore, groups were not perfectly comparable with regard to therapy; for instance, corticosteroids were used more in cases than in controls because of the haematological disease, even if not indicated for the treatment of COVID-19. Nevertheless, to reduce the possible bias of subjects’ selection, such as the different pandemic waves, we performed a strict match for age, sex and period of hospitalization. The level of the inflammatory marker IL-6 was not systematically measured at hospital admission in all the patients and, therefore, we could not evaluate whether it would have an influence on the observed different clinical course of infection between cases and controls. The low absolute number of haematological patients receiving only immunotherapy did not allow us to perform subgroup analyses for the outcome. At last, we enrolled patients before the onset of SARS-CoV-2 vaccination and/or the introduction of monoclonal antibodies in the clinical practice and therefore our findings could not lead to a generalization of the results. Notwithstanding, we were able to describe the singular clinical course of the infection in this special population and our results may represent the starting point for additional studies evaluating if, and how, the course of the disease has changed after the introduction of vaccination or monoclonal antibodies.

In conclusion, an underlying haematological malignancy was associated with a worse clinical outcome in our COVID-19 patients. Furthermore, haematological patients experienced a peculiar clinical course of the infection. This consisted in a higher rate of worsening and, above all, in a longer time to respiratory deterioration compared to non-haematological ones, which independently predicts mortality. Therefore, a careful and prolonged clinical monitoring is needed in this special population.