Characterization of platelet count and platelet indices and their potential role to predict severity in malaria

ABSTRACT

The association of hematological parameters especially platelet parameters with disease severity in malaria is poorly understood. We aimed to characterize the platelet parameters across Plasmodium falciparum and Plasmodium vivax malaria stratified by severity and to elucidate the potential role of platelet parameters to predict disease severity. Individuals > 18 years, of either gender with microscopically proven symptomatic malaria were prospectively enrolled between October 2014 and August 2016 in a tertiary center in Manipal, India. Severity of malaria was defined as per the WHO definition. Among 159 patients, 32 (20.1%) had severe malaria. 116 (73%) had infection with P. vivax, 37 (23%) P. falciparum and 6 mixed infection. Thrombocytopenia was seen in 32 (86.4%) of P. falciparum and 105 (90.5%) of P. vivax malaria cases. Patients with renal failure (p=0.02), shock (p=0.04) and liver dysfunction (p<0.001) had significantly lower platelet count compared to those who did not. Admission platelet count of 50,000 cell/mm³ had a sensitivity and specificity of 65.6% and 70.6% respectively, to discriminate severe malaria. A plateletcrit of 0.05% had a sensitivity and specificity of 65.6 % and of 70.6% respectively. Thrombocytopenia was seen in 89.3% of malaria cases due to both P. vivax and P. falciparum. Platelet count and plateletcrit could be used as markers of disease severity. P. vivax malaria which has been traditionally regarded as ‘benign’ can be as sinister and menacing as P. falciparum malaria and hence warrants equal attention. Unnecessary transfusion of platelets should be avoided.

1. Introduction

Malaria is a serious public health problem in most countries of the tropics. In the year 2015, 89% of the cases of malaria occurring in the South East region were contributed by India [1]. Malaria has a devastating socio-economic impact on developing countries like India considering the disability, mortality and economic loss it causes. Changes in the red blood cells and platelets play a major role in the pathogenesis of complications which can occur in malaria [2,3]. Previous studies have pointed out the utility of platelet count in assessing and predicting the severity of malaria [4,5]. However, some researchers have refuted this claim [6,7]. The data in this regard seem to fall on either side and hence the utility of platelet count as a marker of severity still remains a matter of debate. Although, some studies have evaluated the potential of platelet indices in assessing the disease severity [8,9] the data still seem to be scarce and at a nascent stage. As hematological changes are the hallmark of malaria, it is paramount to explore their utility in determining the severity. Hence, this study was performed to elucidate 1) Hematological parameters across P. falciparum and P. vivax malaria stratified by severity; 2) Prevalence and degree of thrombocytopenia across P. falciparum and P. vivax malaria; 3) Association of platelet parameters with complications in malaria; and 4) Potential role of platelet parameters to predict disease severity.

2. Methods

2.1. Study design and patients’ selection

A prospective cohort study was conducted among microscopically confirmed acute malaria patients aged, ≥18 years, who attended a tertiary care hospital in Manipal, Udupi, India from October 2014 till August 2016. Udupi is situated in the Western Ghats and receives an annual rainfall of more than 4000 mm. It is considered as a low endemicity region for malaria.

Cases with concomitant febrile illnesses, hematological malignancies, platelet disorders, HIV infection, connective tissue disorder, liver diseases and those with mean corpuscular volume <65 fL were excluded. Confirmation of malaria was done by quantitative buffy coat test (QBC). Hematological analysis was done with automated analyzers Coulter HmX from Beckman Coulter at admission. After obtaining informed consent, individual’s history and examination findings were noted. Investigation details such as hematological parameters, renal and liver functions tests, urine analysis, and thin peripheral smears findings were noted in a structured proforma. Details of the treatment received such as anti-malarial regimen, antibiotics, supportive care like blood transfusion, inotropes, inhaled oxygen, dialysis, mechanical ventilation and treatment outcomes were also documented. Attending clinicians prescribed specific antimalarial regimen as per their judgment and the national guideline for the treatment of malaria [10].

2.2. Ethical consideration

This study was conducted after obtaining an approval (IEC 576/2014) from the institutional ethics committee. A written informed consent was obtained from the participants of the study. Patients’ identification and data were anonymized with indirect identification numbers.

2.3. Exposure variables

Platelet parameters including platelet count and platelet indices were the exposure variables. Platelet count was stratified into normal (>150,000 cells/mm³), mild thrombocytopenia (75,000–14,999 cells/mm³⁾, moderate thrombocytopenia (50,000–74,999 cells/mm³), severe thrombocytopenia (25,000–49,999 cells/mm³) and profound thrombocytopenia (<25,000 cells/mm³). Severe malarial anemia was taken as hemoglobin <7g/dl. Leucopenia was defined as WBC < 4000 cells/mm³ and WBC >11,000 cells/mm³ was considered as leucocytosis.

2.4. Outcome variables

Cases were distinguished as either ‘severe’ or ‘non-severe’ malaria as per the definition laid by the WHO [11]. Severity of malaria and individual organ dysfunction were the outcome variables.

2.5. Statistical analysis

Central tendencies were expressed as mean or median (whichever appropriate). Mean±SD was used to summarize continuous variables (Normally distributed) and median (first and third quartile) was used to summarize the variables which were not normally distributed. Frequency and percentage were used to summarize categorical variables. Means between two groups were compared with parametric test viz. independent sample t-test. Anova test was applied for comparison across the groups. Median was compared with non-parametric test viz. Mann Whitney U (between two groups) and Kruskal – Wallis (more than two groups). Chi-square test was used to analyze the qualitative variables. Pearson correlation test was used to analyze the correlation between two variables. Discriminatory efficacy of platelet count and plateletcrit were assessed for complicated malaria using receiver operating characteristic (ROC) analysis. P value <0.05 was taken as significant. Data analysis was done using Statistical Package for the Social Sciences version 22.0 (SPSS, South Asia, Bangalore, India).

3. Results

Of total 159 acute malaria cases studied, 20.1% (32/159) had severe malaria. All severe cases did have complication(s) at their presentation or during the course of hospital stay (shown in Figure 1). 83% (132/159) were males whereas females constituted only 17% of the cases. A comparison of relevant demographic and clinical characteristics between P. vivax, P. falciparum malaria and the mixed infections is given in Table 1. The patients reported at varying intervals from the onset of fever. The median duration of fever in the study cohort was 4 (IQR 3–7) days prior to admission.

Figure 1.

Depicts patients’ selection and outcomes of the study.

Table 1.

Baseline demographic, clinical and hematological characteristics of the study cohort (N = 159).

	P. falciparum (n = 37	P. vivax (n = 116)	Mixed (n = 6)
Mean age, years	33 ± 11.01	38.2 ± 14.7	45 ± 15.8
Gender, male	29	98	5
BMI kg/m2	23.5 ± 4.03	23.5 ± 4	24.3
Days of symptom	4 (6.5,3)	4 (3,7)	5 (3.5,15)
Fever, n (%)	37 (100)	118 (100)	6 (100)
Headache, n (%)	16 (43.2)	40 (34.5)	1 (16.7)
Vomiting, n (%)	9 (24.3)	20 (17.2)	1 (16.7)
Cough, n (%)	1 (2.7)	13 (11.2)	0
Diarrhoea, n (%)	2 (5.4)	0	0
Pallor, n (%)	2 (5.4)	16 (11.8)	2 (33.3)
Icterus, n (%)	9 (24.3)	27 (23.3)	0
Hepatomegaly, n (%)	4 (10.8)	22 (19)	0
Splenomegaly, n (%)	10 (27)	25 (21.6)	1 (16.7)
Hb (gm/dL)	13.79 ± 2.24	12.9 ± 2.04	13.2 ± 1.3
HCT (%)	41.47 ± 6.81	39.0 ± 6.18	39.9 ± 4.9
TLC cell/mm3	5351.±1761	5652 ± 2013	6833 ± 405
PLT (cells/mm3)	72,000 (51,000, 107,000)	62,000 (33,000, 101,200)	71,000 (33,500, 138,250)
MPV (fL)	9.78 ± 1.31	9.51 ± 1.36	9.53 ± 1.5
PCT (%)	.07 (.04, 0.1)	.06 (0.03, 0.09)	.09 (.06, 0.1)
PDW (%)	17.02 ± .84	17.3 ± .98	17.5 ± .77
	n = 10	n = 34	n = 2
PT/INR (in sec)	1.13 ± .077	1.13 ± .09	1.2 ± 0.07
APTT (in sec)	32.08 ± 2.4	32.4 ± 4.2	33.8 ± 1.1

Hb: Hemoglobin, HCT: Hematocrit, TLC: Total leucocyte count, PLT: Platelet count, MPV: Mean platelet volume, PCT: Plateletcrit, PDW: Platelet distribution width.

3.1. Hematological parameters at presentation across species

In falciparum cohort, 3.4% males and 12.5% females had hemoglobin less than 10g/dL. In vivax cohort, 6.1% males and 11.1% females had hemoglobin less than 10 g/dL (Additional file 1). Notably, no individual had severe anemia at presentation. During the course of hospital stay, five patients developed severe anemia. Leucopenia was seen in 24.3% and 22.4% of the falciparum and vivax cohort, respectively. Leukocytosis (WBC>11,000 cells/mm3) was seen in one patient with vivax malaria and none with falciparum malaria

Thrombocytopenia was noted in 89.3% (142/159) patients. Species-wise, 86.4% (32/37) of P. falciparum patients and 90.5% (105/116) of P. vivax patients had thrombocytopenia. Among mixed infection, 83.3% (5/6) patients had thrombocytopenia. Normal platelet count was seen in 10.7% (17/159) patients. There was no association or trend with fever duration and platelet count, either in P. vivax malaria or P. falciparum malaria. (Additional File 2a & 2b). However, 88% (128/144) of patients with thrombocytopenia had fever ≤7 days. The median duration of fever in thrombocytopenia patients is 4 (IQR 3–6) days, implying that thrombocytopenia is an early event. Platelet indices and their distribution across species (and by disease severity) have been shown in Tables 1 & 2. Distribution of degree of thrombocytopenia is depicted in Table 3. Twenty (17.4%) and twenty-eight (24.3%) patients in P. vivax cohort had profound and severe thrombocytopenia compared to one (2.7%) and seven (18.9%) patients in P. falciparum cohort, respectively.

Table 2.

Hematological parameters across the entire cohort and species at presentation stratified by severity.

	Entire cohort (N = 159)			Plasmodium falciparum (N = 37)		Plasmodium vivax (N = 116)
	Non-severe 127 (79.9%)	Severe 32 (20.1%)	p-value*	Non-severe 33 (89.2%)	Severe (10.8%)	Non-severe 89 (76.7%)	Severe 27 (23.3%)	p-value**
HB (g/dL)	13.35 ± 2.02	12.38 ± 2.19	0.01	13.96 ± 2.04	12.42 ± 3.64	13.10 ± 2.02	12.2 ± 1.98	0.06
HCT (%)	40.30 ± 6.11	37.18 ± 6.7	0.01	42 ± 6.10	37.17 ± 11.46	39.76 ± 6.13	36.837 ± 5.91	0.03
TLC (cells/mm3)	5558 ± 1809	5900 ± 2876	0.4	5106 ± 1588	7375 ± 2046	5743 ± 1893	5351 ± 2384	0.3
PLT (x103cells/mm3)	74 (41, 107)	30 (19, 81)	0.001	79 (52,116)	49 (29, 82)	74 (41, 107)	25 (16, 76)	0.001
MPV (fL)	9.4 ± 1.33	9.91 ± 1.22	0.09	9.582 ± 1.13	11.45 ± 1.69	9.4 ± 1.40	9.79 ± 1.21	0.2
PCT (%)	0.06 (0.04, 0.09)	0.02 (0.01, 0.07)	0.001	0.07 (0.04,0.1)	0.06 (0.031, 0.1)	0.06 (0.04, 0.09)	0.023 (0.01, 0.06)	0.001
PDW (%)	17.2 ± 0.99	17.32 ± 0.80	0.8	17.01 ± 0.89	17.07 ± 0.22	17.36 ± 1.02	17.37 ± 0.862	0.9

* p-Value is for comparison of Non-severe vs Severe malaria across the whole cohort

** p-Value is for comparison of Non-severe vs Severe P. vivax cohort

Table 3.

Distribution of thrombocytopenia with its grading across the species.

Degree of thrombocytopenia	Grading	Platelet count (cells/mm3)	P. falciparum (n = 37) n (%)	P. vivax (n = 116) n (%)	Mixed infection (n = 6) n (%)
Profound	Grade 4	<25,000	1 (2.7)	20 (17.4)	1 (16.7)
Severe	Grade 3	25,000–49,999	7 (18.9)	28 (24.3)	1 (16.7)
Moderate	Grade 2	50,000–74,999	11 (29.7)	15 (13.0)	1 (16.7)
Mild	Grade 1	75,000–149,999	13 (35.1)	41 (35.7)	2 (33.3)
Normal	Grade 0	150,000 & above	5 (13.5)	11 (9.6)	1 (16.7)

A positive correlation was found between platelet count and plateletcrit (r = 0.4; p-value <0.001) platelet count and platelet distribution width had a negative correlation (r = −0.5; p < 0.001). No significant correlation was found between platelet count and platelet distribution width (Additional file 4).

Table 4 shows the association of platelet count and platelet indices with individual complications Patients with renal failure (p = 0.02), shock (p = 0.04) and liver dysfunction (p < 0.001) had significantly lower platelet count compared to those who did not. Notably, platelet count was not associated with spontaneous bleeding. Patients who had spontaneous bleeding had significantly lower mean platelet volume (MPV) (p = 0.02) as compared to those with no bleeding. Difference between the MPV in patients with renal failure as compared to patients with normal renal function was also statistically significant (p = 0.01).

Table 4.

Comparison of median platelet count and mean platelet volume with respect to complications.

Complication	No. of patients (N = 32)	Median platelet count (X 103cells/mm)(IQR*) with this complication	Median platelet count (x 103cells/mm3)(IQR) without this complication	p-value	Mean platelet volume (fl) with this complication	Mean platelet volume (fl) without this complication	p-value
Cerebral malaria	3 (9.3%)	88 (37,128)	63.5 (36,103)	1	9.8 ± 1.3	9.5 ± 1.3	0.78
Shock or circulatory collapse	6 (18.75%)	17 (10,25)	71 (39,104)	0.04	9.8 ± .45	9.5 ± 1.3	0.33
Severe acidosis	3 (9.3%)	37 (8,128)	66 (37.5,103)	1	9.1 ± .25	9.6 ± 1.3	0.60
Pulmonary edema/ARDS	3 (9.3%)	37 (14,90)	66 (37.5,104)	1	10.3 ± 1.99	9.5 ± 1.3	0.36
Severe malarial anemia	5 (15.6%)	88 (24,178)	65 (37,102)	1	8.9 ± 2.1	9.6 ± 1.2	0.27
Spontaneous bleeding	5 (15.6%)	29 (13,37)	70 (39,104)	0.07	9.1 ± .28	9.6 ± 1.3	0.02
Renal failure	13 (40.6%)	28 (16,48)	72 (39,104)	0.02	10.4 ± .85	9.5.±1.3	0.01
Clinical evidence of jaundice and vital organ dysfunction	18 (56.25%)	23.5 (16,37)	72.5 (41,107)	<0.001	9.9 ± .96	9.5 ± 1.3	0.18
Death	1 (3.125%)	37	66 (37,103)	1	9.2	9.±1.3	0.78

3.2. Discriminatory performance of platelet count and plateletcrit with respect to severity

In the ROC analysis, the Area under the curve (AUC) was 71.3 for platelet count with respect to disease severity. Further analysis showed that a platelet count of 50,000 cell/mm³ had a sensitivity of 65.6% and a specificity of 70.6% to discriminate severe malaria. Also, AUC was 71.8 for plateletcrit with respect to disease severity. A plateletcrit of 0.05% had a sensitivity of 65.6% and a specificity of 70.6% to discriminate severe malaria (Figure 2).

Figure 2.

ROC and area under the curve (AUC) of platelet count and plateletcrit with respect to severity.

3.3. Patients’ treatment, outcomes and supportive requirement

All the patients of falciparum malaria were treated with ACT-based therapy. Combination of artemether and lumefantrine was given in 21.6% (8/37) patients and 78.37%(29/37) were administered intravenous artesunate and doxycycline. In the vivax cohort, 44.8% (52/116) patients received ACT-based therapy, while the rest were treated with chloroquine. All the individuals with P. vivax malaria received 14 days of primaquine after ruling our G6PD deficiency. One patient of vivax malaria died among the study participants. Twelve patients received ICU care, 13 patients received blood products, 6 patients required inotropes and one was mechanically ventilated.

4. Discussion

Our study confirms the fact that thrombocytopenia is prevalent in malaria. Majority of the patients (89.3%) had thrombocytopenia in our study. Among the plasmodia species, 86.4% P. falciparum malaria patients had thrombocytopenia, the frequency was astonishingly high (90.6%) in P. vivax malaria. Similar rates of thrombocytopenia have been reported in other studies [8,12,13]. The incidence of thrombocytopenia in a literature review has been reported as 24–94% [14]. Mechanism of thrombocytopenia is not understood well and multiple mechanisms such as immune mediated, platelet phagocytosis, increased oxidative stress, platelet aggregation, etc., are thought to be responsible for the same.

4.1. Severity of thrombocytopenia

The median platelet counts at admission in falciparum, vivax and mixed infection were 72,000 cells/mm³, 62,000 cells/mm³, and 71,000 cells/mm³ respectively. Majority of the patients had mild thrombocytopenia (35.1% in P. falciparum and 35.7% in P. vivax). Surprisingly, in our cohort we found that the severity of thrombocytopenia was much more in vivax malaria as compared to falciparum malaria. Severe and profound thrombocytopenia in the vivax cohort was seen in 24.3% and 17.4% patients where as in falciparum it was seen in 18.9% and 2.7% of the patients, respectively. This is in contrast to the usual belief that falciparum malaria causes more severe thrombocytopenia as shown by previous studies [8,13,15]. Studies have shown that there is a stronger host response in vivax malaria in terms of cytokines released as compared to falciparum malaria [16]. This increased cytokine production may be responsible for profound thrombocytopenia in vivax malaria. Another explanation could be the emergence of a new genotype of P. vivax which is more virulent. However, further studies are required with vivax malaria patients to fill the lacuna in the knowledge in this field. Nevertheless, it underlines the importance of recognizing P. vivax as a cause of severe malaria.

4.2. Predictive performance of platelet count for severity

Notably, platelet count at admission was significantly lower in severe cases as compared to non- severe cases (74,000 cells/mm³ to 25,000 cells/mm³ (p = <0.001)) which is in unison with other studies [17,18]. In ROC analysis, we found that the area under the curve (AUC) was 71.3 for platelet count to discriminate severity. A platelet count 50,000 cells/mm³ had a sensitivity of 65.6% and a specificity of 70.6% to discriminate severe malaria. Similarly, plateletcrit of 0.05% had a specificity of 71.2% and a sensitivity of 65.6%. Rao et al. showed that a person with platelet count of less than 50,000 cells/mm³ was 8.5 times more likely to have complicated malaria [19]. In an Indonesian study authors proposed a platelet count of ≤20,000 cells/mm³ could be used as a criterion for severe malaria [20]. A Previous study from India has suggested that platelet count can be used as a prognostic marker and could predict the severity of malaria. In a tertiary care hospital-based study authors identified severe thrombocytopenia and leukocytosis to be having independent association with prolonged hospitalization, intensive care requirement and mortality [4]. However, the utility of thrombocytopenia as a prediction tool for disease severity has been a matter of debate. A large retrospective analysis by Hanson et al. showed that admission platelet count correlated inversely with disease severity. Also, the admission median platelet count (IQR) in the patients who died was 30 (22 to 52) ×10⁹/L compared to 50 (34 to 78) × 10⁹/L in survivors (p = 0.0001). However, they concluded that the platelet count alone as a prognostic marker had limited value as compared to other laboratory indices such as lactate, base deficit, plasma creatinine total bilirubin and bedside clinical indices such as GCS and Respiratory based Coma Acidosis Malaria (RCAM) Score [6]. Nonetheless, we urge that is platelet count is a simple investigation which is available at most health centers as compared to lactate and base deficit which are available only in sophisticated laboratories and are not accessible to vast majority of individuals with malaria. More studies with a larger cohort of severe malaria are required to validate our findings and to identify a cut off platelet value which could guide and help the clinicians in triaging the patients and thereby help in better and cost-effective management. Further, it is not clear whether the platelets play a role in the pathogenesis of severe malaria or are they innocent by-standers, the elucidation of which might open new avenues of therapy.

4.3. Predictive performance of platelet indices for severity

The data about the alterations in platelet indices in malaria remains scarce hence in this study we analyzed the effect of malaria on the platelet indices such as mean platelet volume (MPV), plateletcrit (PCT) and platelet distribution width (PDW). The median plateletcrit for complicated cohort was 0.023% as compared to 0.06% in uncomplicated cohort (p = <0.001). No significant difference was found in MPV and PDW among the cohorts. Platelet count and its indices were studied in patients who had complications to those who did not. Those patients who had shock, renal failure, and liver dysfunction along with vital organ dysfunction had a significantly lower mean platelet count at admission (Table 4). Saravu et al. reported the admission platelet count in patients with renal failure was significantly lower as compared to those who did not have renal failure [8]. A large retrospective study in Colombia found significant differences in the platelet count and PCT among all the complications. This could have been due to the fact that their criteria for the complicated malaria were more liberal and according to their local guidelines which included both major and minor criteria [21]. They also found higher PDW values (in patient with cerebral malaria and renal complications) compared with the patients who did not have complications, a finding which was different from our study as we did not find any significant difference. In the present study, there was inverse correlation between platelet count and MPV (r = 0.5; p = <0.001) which was in accordance with other studies [22,23]. When platelets decrease in number there occurs an early release of platelets from the bone marrow. These early released platelets are larger in size. Also, the stimulation of megakaryocytes in the marrow by thrombopoetin in response to low platelets produces larger platelets as their nucleus becomes hyper lobulated with a higher deoxyribonucleic acid (DNA) content [21].

In our study, five patients had spontaneous bleeding. Bleeding is not very common in malaria because of the larger platelets released into the circulation which compensate for the low platelet count. These giant platelets maintain the primary hemostasis and hence prevent the bleeding even with severe thrombocytopenia. The patients who bled had a MPV of 9.2, 9.5, 8.7, 9.1, 9.2 fl as compared to 9.6 fl, in those who did not bleed (p = 0.02). Even though the platelet count was lower in those who bled it was not statistically significant. This is in line with the explanation as to why bleeding is a rare phenomenon in malaria and those who bleed have a smaller volume of platelets. Hence, unnecessary transfusion of platelets by treating physician should be avoided. Similar kind of sentiment has been echoed by other researchers [6]. Also, it can alert the clinician if the MPV does not increase sufficiently, the patient may be at risk of bleeding.

Patients, in our study, with renal failure had significantly higher MPV (10.4fl (±0.85) to 9.5fl (±1.3); p = 0.01). MPV is considered a potential marker of platelet reactivity and is associated with proinflammatory disease states. Studies have shown an association of increased MPV with renal dysfunction [24,25]. It is seen that the life-span of platelets in uremia is decreased which leads to platelet production and release of young platelet (of higher MPV) from the marrow.

This study has characterized platelet count and platelet indices in a setting with low endemicity for malaria where both P. vivax and P. falciparum infections are reported. Our study certainly has limitations. The test of significance could not be applied on the falciparum cohort due to inadequate numbers. Some of the severity criteria like level of parasitemia were not available as it was not routinely performed. The speciation was based on microscopy and not confirmed by PCR. The study was conducted in hospitalized malaria patients; hence the frequency of severity and thrombocytopenia could be higher. Due to financial constraints, patients’ platelet counts were not followed up systematically until recovery.

5. Conclusions

Thrombocytopenia is predominant in both P. falciparum and P. vivax malaria. P. vivax malaria is associated with profound thrombocytopenia compared to P. falciparum malaria. Present study brings out the potential utility of platelet count and plateletcrit in identifying severe malaria. Complications such as shock, renal failure and liver dysfunction are associated with lower platelet count and plateletcrit at admission. Further, P. vivax malaria which has been traditionally regarded as ‘benign’ malaria warrants more attention both by the clinicians and researchers. Since platelet count is easily available in most primary health centers in resource-constrained settings, further studies are needed to validate utility of platelet count as a severity marker for better triaging in patient management. Since clinical bleeding in malaria is not common, unnecessary transfusion of platelets is uncalled for and should be refrained from. Future studies must explore the role of platelets in the pathogenesis of severe malaria.

Funding Statement

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors

Authors’ contributions

PG and KS: conception and design; PG and KS: acquisition of data; PG, VG and KS: analysis and interpretation of data; PG, VG and KS: drafting the manuscript or revising it critically for important intellectual content; PG, VG and KS: final approval of the version to be published and agree to be accountable for all aspects of the work.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplementary data can be accessed here