Temporal changes of haematological and radiological findings of the COVID-19 infection—a review of literature

In early December 2019, a series of pneumonia of unknown causes with clinical features that resemble viral pneumonia were identified in Wuhan, Hubei, China [1]. The World Health Organisation (WHO) officially named the clinical condition COVID-19 (coronavirus disease-19) [2] and the Coronavirus Study Group of the International Committee on Taxonomy of Viruses renamed the virus “severe acute respiratory syndrome coronavirus 2” (SARS-CoV-2) [3]. Coronaviruses belong to the family Coronaviridae and the order Nidovarales, a family that includes viruses that cause diseases ranging from the common cold to severe acute respiratory syndrome (SARS) and the Middle East respiratory syndrome (MERS) [4]. WHO has since characterised COVID-19 as a global pandemic [5]. The WHO Coronavirus Disease (COVID-19) Dashboard reflects that 220 countries have been affected, with 58,425,681 confirmed cases and 1,385,218 deaths globally as of 23 November 2020.

The incubation period for COVID-19 was estimated to be 5.1 days, and 97.5% became symptomatic within 11.5 days [6]. The main clinical symptoms include fever, cough, fatigue, sputum production, shortness of breath and myalgia/arthralgia [1]. Other minor symptoms include headache or dizziness, diarrhoea, nausea and vomiting [7]. Major complications of patients with COVID-19 include acute respiratory distress syndrome (ARDS) and some also progress to multi-organ failure [8]. The severity of patients with COVID-19 runs on a spectrum from mild, severe to critical [9].

Given the wide spectrum of severity that can be found in a patient with COVID-19, it is important to identify potential clinical characteristics that would help predict the clinical outcome early. This in turn could guide management in terms of resource allocation. Thus far, several potential predictors of outcome have been suggested to help monitor patients which may develop severe complications. Lymphopenia has been shown to be a prominent laboratory finding in severe COVID-19 patients and is important in predicting the prognosis [10, 11]. Liu et al. [12] has also reported that neutrophil-to-lymphocyte ratio is an independent risk factor for in-hospital mortality of patients. Other potential haematological predictors of outcome that have been proposed included C-reactive proteins (CRP) levels [13], lactate dehydrogenase (LDH) levels [14], cardiac troponin-I [15], low-density lipoproteins [16]. In a recently published study, Liang et al. [17] proposed a new clinical risk score to predict the occurrence of critical illness in hospitalised COVID-19 patients. The risk score included many of the above-mentioned laboratory markers as well as chest radiographic abnormality. It may also be fruitful to look at the temporal progression of blood and radiological results whilst comparing between patient groups with a mild disease course and those with a severe outcome or even death. Insights drawn can demonstrate typical progressions of blood and radiological results which may help to identify patients at high risk of severe outcomes.

It should also be recognised that a significant proportion of infected patients are asymptomatic in the early days after documented exposure, with an estimated proportion to be about 18% [18]. In fact, higher proportions of asymptomatic cases were reported in some studies with 87.8% of the patients who were tested positive being asymptomatic [19]. The concern lies in the fact that though asymptomatic, these carriers are still very capable of transmitting COVID-19 via person-to-person contact, thereby propagating the pandemic [20]. Therefore, being able to identify such patients at an early stage via haematological or radiological investigations would be beneficial, especially in regions with limited availability of and accessibility to COVID-19 test kits.

A systematic literature review was conducted using the online databases, Pubmed, Embase and Cochrane databases. The search was done with the aim of identifying publications that reported the radiological or blood changes in patients with COVID-19.

This was done in accordance with the Cochrane Handbook of Systematic Reviews and Meta-analysis and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines [21]. The electronic searches covered the period from 1 December 2019 to 29th March 2020. They were done using a combination of Medical Subject Headings (MeSH) and non-MeSH key terms with Boolean operators (Additional files 1 to 6).

All searches underwent double-blind screening of the title and abstract by two researchers for inclusion and exclusion criteria. The inclusion criteria are studies that include patients diagnosed with COVID-19 and report radiological (computed tomography (CT) of thorax/chest radiograph) or blood (full blood count) findings together with their date since onset of symptoms when the respective investigations were done. Case reports, studies with fewer than 10 patients, abstracts, letters to editors, editorials, commentaries, features, news, guidelines, opinion pieces and non-English language papers were excluded. Duplicates were removed, and the selected articles underwent a full-text review to verify quality and eligibility. Conflicts were resolved by consensus.

Data was extracted from text, tables and figures in each selected paper using a standardised method into a preformatted database by a single researcher. All extracted data was then verified by a second researcher independently. A sample of the extraction grid is included in Additional file 2.

Overall, 806 articles were identified through database searching. After screening the titles and abstract, 689 articles were excluded according to the exclusion criteria leaving 117 articles for full-text review. A further 100 articles were excluded after full text review and 27 articles remained to be included. A summary of the study selection process is shown in the PRISMA flow diagram (Fig. 1) [12, 22‐44].

Temporal changes of haematological findings of the COVID-19 infection was estimated using their mean and SD for the corresponding data of onset. If mean and SD were not available, it was estimated those using median and quartiles [45]. When there are two are more mean readings from different studies for the same date of onset (DOO), a weighted average was computed and used for generating trend. Trend for the blood parameters by DOO were generated, only if the parameters were available for at least 3 DOO time points.

In patients diagnosed with COVID-19, laboratory findings that were commonly present included lymphopenia, raised CRP levels and raised LDH levels. By comparing the peripheral blood investigation findings between the earlier phase (DOO ≤ 7 days) and the later phase (DOO > 7 days) of the disease, there was more prominent lymphopenia and raised inflammatory markers like CRP and LDH levels in the latter. Neutrophil counts were higher in the later phase compared to the earlier phase as well.

Regardless of the number of days since onset of symptoms, lymphopenia can be seen in most hospitalised patients with COVID-19. Besides studies from China, studies from Saudi Arabia [48], Iran [49], India [50], South Korea [51], Singapore [52] and cruise ship (Diamond princess) [53] have also confirmed this finding However, the extent of lymphopenia is more prominent in the early phase (DOO ≤ 7 days) compared to those in the later phase (DOO > 7 days). It has been hypothesized that lymphopenia could be due to direct infection of the lymphocyte by the virus, destruction of lymphatic organs, lymphocyte apoptosis due to inflammation or inhibition of lymphocytes due to metabolic disorders like lactic acidosis [54, 55]. As such, we postulate that at a later timing as the disease progresses, more severe metabolic processes could further inhibit lymphocytes. Furthermore, worsening lymphopenia is also evident in the clinical course of patients with severe outcomes compared to those without. As such, the progression of lymphopenia in COVID-19 patients would be a good predictor of severity and outcomes. Since disease progression is associated with worsening lymphopenia and hence severity, lymphocyte counts could be employed as a tool for prognostication in terms of the severity of the disease especially for patients in the later phase of the disease [11].

Elevations in biomarkers such as CRP and LDH levels were also more significant in the later phase of the disease (DOO > 7 days) than the earlier phase (DOO ≤ 7 days). Similar elevation in CRP was also observed among patients in Saudi Arabia [48] and Iran [49] and was associated with mortality. CRP, an acute phase reactant, increases rapidly after the onset of inflammation, cell damage or organ injury in response to inflammatory cytokines [56]. Similarly, LDH is released from cells upon damage of their cytoplasmic membrane [57]. As such, a pulmonary disease such as COVID-19 would cause an increase in these biomarkers. At a later phase, even higher levels of CRP and LDH could be a manifestation of disease progression as a result of the organ damage due to direct attacks from the virus causing excessive inflammatory response. Given that these biomarkers correlate positively to the disease progression, it may also be useful to monitor severity in patients.

Neutrophil counts in the later phase (DOO > 7 days) were also noted to be higher than those in the earlier phase (DOO ≤ 7 days). Similar phenomenon was observed in Singapore [52] and Iran [49]. The increase in neutrophil count as the disease progresses was thought to be due to the sustained virus invasion leading to a cytokine storm [32]. Furthermore, patients with severe outcomes have a more prominent elevation in neutrophil counts than those without severe outcome. Thus, an increase in neutrophil count could be associated with disease progression and severity. This would be beneficial as neutrophil counts could then be used to monitor the severity and disease progression in patients with COVID-19.

Overall, the most commonly reported lung feature from our analysis is GGO (1056/1610 scans, 65.6%), followed by GGO with consolidation (135/293 scans, 46.1%), then consolidation (462/1325 scans, 34.9%), and finally crazy paving pattern (280/1064 scans, 26.3%). Similar patterns were observed in studies elsewhere [50, 53, 58‐60]. These CT findings are likely related to the complex pathological changes in the lungs of patients with COVID-19. Tian et al. [61] reported that histologically, the main findings in the lungs include injury to the alveolar epithelial cells, hyaline membrane formation, and hyperplasia of type II pneumocytes, all of which are components of diffuse alveolar damage. Consolidation by fibroblastic proliferation with extracellular matrix and fibrin forming clusters in airspaces is evident. Zhe et al. [62] also concurred, reporting that there is diffuse alveolar damage with cellular fibromyxoid exudates, as well as desquamation of pneumocytes and hyaline membrane formation, indicating acute respiratory distress syndrome (ARDS). These pathological changes may be the main pathological basis of the CT findings.

From our analysis, the frequency of GGO was higher in early-phase disease than in later-phase disease. Pathologic examinations reported by Tian et al. [61] found that edema, proteinaceous exudate, focal reactive hyperplasia of pneumocytes with patchy inflammatory cellular infiltration, and multinucleated giant cells were all present in the early phase of the disease. However, in later-phase disease, the frequency of consolidation and that for crazy paving pattern were higher than those in early-phase disease. We postulate that this might be due to further infiltration of the lung parenchyma and they are also typical features of the underlying pathophysiology of an organising pneumonia [63]. Also, consolidation is considered as an indication of disease progression and crazy paving pattern could be the signal of COVID-19 entering progressive or peak stage over time [64].

In terms of lesion distribution, the SARS-CoV-2 virus seems to cause a significantly higher frequency of bilateral lung involvement compared to unilateral, and this is even more so in the later phase. Multilobe involvement also seems to be a prominent feature, with 4–5 lobe involvement occurring at the highest frequency. This pattern is further evident in the later phase. Overall, the lesions seem to be mostly distributed peripherally and this is also the case in early-phase disease. However, in later-phase disease, the lesions appear to be distributed both peripherally and centrally with highest frequency. This predominantly bilateral, multilobar and peripheral lung distribution concurs with other radiological studies [65] on the virus and appears to be a hallmark chest CT feature [65]. This may suggest a high virulence factor of the SARS-CoV-2 virus due to its extensive penetration of the lung parenchyma [66].

All these temporal changes on chest CT imaging suggest the utility of this investigation modality in monitoring for disease progression in COVID-19 patients, and this is in agreement with the consensus statement from the Fleischner Society [67].

Our analysis has also demonstrated key differences in chest CT findings between groups with severe outcomes and those with non-severe outcomes. The former typically exhibits a greater degree and frequency of consolidation, more extensive lesion distribution, and a higher frequency of multilobe bilateral lung involvement. Temporally, the lung involvement on CT of the former also progressed more extensively and rapidly, while no significant changes were noted over time for the groups with non-severe outcomes. Additionally, the worsening radiological features coincide with the worsening of lymphopenia, neutrophilia, CRP elevation and LDH elevation which signify the worsening of systemic inflammation and viral invasion. These all suggest that such CT findings could help predict and identify patients at risk of severe clinical outcomes.

This review is included 27 studies, all of which were conducted in China. Also, amongst the included studies, only one study was conducted among critical ill patients. This is a limitation as the number of COVID-19 cases and deaths in many other countries including the United States and several European countries have since exceeded that of China [7]. However, our review findings are similar to emerging evidence of haematological and radiological from other parts of the world [48‐53, 58‐60, 68].

Majority of the studies (n = 25) were retrospective and thus had inherent design-specific limitations such as bias in selection, residual confounding, and temporal ambiguity. Some studies were unable to be included in the evaluation of infection time course due to the lack of precise date of symptom onset reported. Furthermore, the number of studies that reported findings on COVID-19 patients in the later phase (DOO > 7 days) was also deficient. In addition, there were few studies included in this review that monitored the clinical course of a patient from symptom onset to their ultimate outcome, which would allow for better analysis of clinical progression of either the peripheral blood investigations or radiological investigations. Also, different normal ranges for the peripheral blood investigation results were reported in different studies which made comparison difficult, particularly when trying to account for slight elevations or decreases in the results, which at an early phase may herald disease progression. Furthermore, this review did not account for the treatments administered to the patients in the various studies, which include antivirals, corticosteroids, antimicrobials and oxygen therapy among others. Although there is currently still no reliable and effective treatment for COVID-19 to date, the various medical interventions may have altered the natural progression of the disease. Hence, the temporal changes of both the peripheral blood and CT imaging findings may not be the most reliable and applicable to other patient populations.

As such, further reviews can be conducted to address the above limitations of this review, particularly because many studies on COVID-19 patients in countries besides China have since been published.

More studies in the outpatient, primary care, or community settings are also needed to get a full picture of the spectrum of clinical severity. Additional effort can be also made to conduct a similar study in other patient populations, such as the paediatric and adolescent age groups. Further studies can be done to distinguish between the laboratory and radiological manifestations of COVID-19 and the rest of its viral family, particularly SARS and MERS, as well as with other kinds of viral pneumonia such as influenza virus, parainfluenza virus, adenovirus, respiratory syncytial virus, rhinovirus, human metapneumovirus, and mycoplasma pneumonia.

With time progression, peripheral blood results reflect worsening lymphopenia, more significant elevations in CRP and LDH as well as that for neutrophil counts. Temporal change also shows decrement of the frequency of GGO, as opposed to increasing consolidation and crazy paving patterns on chest CT imaging. The lesion distribution in early-phase disease already appears extensive and continues causing more lung involvement over time. These findings suggest a positive correlation between temporal progression, disease severity and the reported blood and radiology results. Awareness of these trends in blood results and imaging changes may assist clinicians in making ICU admission decisions when resources are limited.