Abstract
Background:
Inflammation has been suggested to play a role in heart failure (HF) pathogenesis. However, the role of platelet-to-lymphocyte ratio (PLR), as a novel biomarker, to assess HF prognosis needs to be investigated. We sought to evaluate the impact of PLR on HF clinical outcomes.
Methods:
English-published records in PubMed/Medline, Scopus, and Web-of-science databases were screened until December 2023. Relevant articles evaluated PLR with clinical outcomes (including mortality, rehospitalization, HF worsening, and HF detection) were recruited, with PLR difference analysis based on death/survival status in total and HF with reduced ejection fraction (HFrEF) patients.
Results:
In total, 21 articles (n = 13,924) were selected. The total mean age was 70.36 ± 12.88 years (males: 61.72%). Mean PLR was 165.54 [95% confidence interval (CI): 154.69–176.38]. In total, 18 articles (n = 10,084) reported mortality [either follow-up (PLR: 162.55, 95% CI: 149.35–175.75) or in-hospital (PLR: 192.83, 95% CI: 150.06–235.61) death rate] and the mean PLR was 166.68 (95% CI: 154.87–178.50). Further analysis revealed PLR was significantly lower in survived HF patients rather than deceased group (152.34, 95% CI: 134.01–170.68 versus 194.73, 95% CI: 175.60–213.85, standard mean difference: −0.592, 95% CI: −0.857 to −0.326, p < 0.001). A similar trend was observed for HFrEF patients. PLR failed to show any association with mortality risk (hazard ratio: 1.02, 95% CI: 0.99–1.05, p = 0.289). Analysis of other aforementioned outcomes was not possible due to the presence of few studies of interest.
Conclusion:
PLR should be used with caution for prognosis assessment in HF sufferers and other studies are necessary to explore the exact association.
Plain language summary
Inflammation plays a role in heart failure (HF), and a blood test called the platelet-to-lymphocyte ratio (PLR) might be helpful in predicting patients’ outcomes. We found that deceased HF patients had higher PLR values in comparison to those who survived, irrespective of cardiac pump function, with similar pattern for patients with decreased cardiac function (HF with reduced ejection fraction). However, this biomarker failed to show any significant association with death risk. In conclusion, PLR may have some potential to help predict HF prognosis, but it needs more research and physicians should probably be cautious about using PLR alone in clinical settings.
Introduction
Heart failure (HF) is officially defined as ‘a complex clinical syndrome that results from any structural or functional impairment of ventricular filling or ejection of blood. The cardinal manifestations of HF are dyspnea and fatigue, which may limit exercise tolerance, and fluid retention, which may lead to pulmonary and/or splanchnic congestion and/or peripheral edema’. 1 Its prevalence ranges from 0.2% to 17.7%. 2 More than 5.5 million individuals suffer from HF in the United States and this number rises to 23 million patients around the globe. 3 Elderly people are more susceptible to HF in a way that about 10% of people aged at least 70 years experienced HF. 4 Despite recent progress in HF diagnosis and introduction of some novel therapies, this disease is not well controlled.5,6 Mortality is another dilemma in HF and it has been reported to be 50–75% in the first 5 years after HF diagnosis.7–9 One-year HF case fatality rate has been reported from 4% to 45%. 2 Moreover, HF economic burden should also be considered. Current reports indicate that 1–3% of overall health expenditure is attributed to HF management, mostly related to repetitive hospitalizations as well as extended length of hospital stay. 10 The annual economic cost of HF was reported to be approximately $108 billion, necessitating early assessment to decrease unfavorable outcomes. 11 Till now, several HF prognostic factors have been suggested including shock index, pro-brain natriuretic peptide, blood pressure, age, and hemoglobin.12–16
Inflammation, thrombosis, and inflammatory markers have been recently reported to play a role in pathogenesis cardiovascular diseases (CVDs), especially HF.17–21 Platelets as well as leukocytes are major inflammatory mediators and the division of platelets to lymphocytes results in introduction of a new inflammatory index, named platelet-to-lymphocyte ratio (PLR). 22 This index has been reported to have practical utility in a wide spectrum of disorders like renal disorders, CVDs, and malignancies.23–25 Although this simple tool has been used to assess prognosis in HF sufferers, reported outcomes are still controversial in a way that some records were in favor of its usefulness and the others failed to find any prognostic capability.26–28
With respect to these inconsistent results, we sought to implement this systematic review and meta-analysis to assess probable impact of PLR on clinical outcomes among individuals with HF.
Materials and methods
Protocol registration
This systematic review and meta-analysis was performed in context of Preferred Reporting Items for Systematic Reviews and Meta-Analyses. 29 We also registered this study in the International Prospective Register of Systematic Reviews (PROSPERO) electronic database (CRD42022335383).
Eligibility criteria
All English-published peer-reviewed cross-sectional, cohort, case control, systematic reviews, and randomized clinical trial (RCT) studies reported the PLR effect with clinical outcomes in HF subjects were recruited. In addition to animal studies, case report and case series studies as well as letters without providing the desired output or with incomplete data were excluded.
Search strategy
We searched PubMed/Medline, Scopus, and Web of science electronic databases without any time limitation. All fields were searched in PubMed/Medline and Web of science databases. In Scopus, title, abstract, and keywords were investigated. We used the following search strategy in all aforementioned databases till December 2023: (‘heart failure’ OR ‘cardiac failure’ OR ‘heart insufficiency’ OR ‘cardiac insufficiency’ OR ‘congestive heart failure’ OR ‘congestive cardiac failure’ OR ‘decompensated heart failure’ OR ‘decompensated cardiac failure’ OR ‘decompensated heart insufficiency’ OR ‘decompensated cardiac insufficiency’ OR ‘acute decompensated heart failure’ OR ‘acute decompensated cardiac failure’ OR ‘acute decompensated heart insufficiency’ OR ‘acute decompensated cardiac insufficiency’ OR ‘hf’) AND (‘platelet★ to lymphocyte★ ratio’ OR ‘platelet★-lymphocyte★’ OR ‘platelet★-lymphocyte★ ratio’ OR ‘platelet★ to lymphocyte★’ OR ‘platelet★-to-lymphocyte★ ratio’ OR ‘platelet★-to lymphocyte★ ratio’ OR ‘platelet★ to-lymphocyte★ ratio’ OR ‘platelet★/lymphocyte★ ratio’ OR ‘platelet★/lymphocyte★’ OR ‘plt/lymph’ OR ‘plt to lymph ★ ratio’ OR ‘plt to lymph ratio’ OR ‘plr’).
Data management and selection process
Titles and abstracts were carefully screened by two independent reviewers and all relevant records with their full texts were gathered. In case of any duplicated articles, only single record was considered. Figure 1 illustrates flow diagram of current study.
Flow diagram of the study.
Data collection process
In each record, the following items were extracted: first author’s name, publication year, study design, sample size, male gander, age [mean ± standard deviation (SD) or median (interquartile range (IQR)), as reported], follow-up duration (as applicable), PLR [mean ± SD, median (IQR), range, as reported], PLR quartiles, tertiles and cutoff values (as reported), and outcomes (follow-up death, in-hospital mortality, rehospitalization, worsening of HF, and HF detection, as reported). Any disagreement was resolved by consensus.
Risk of bias and quality assessment
The quality of cross-sectional studies was assessed through a critical appraisal tool (AXIS). 30 Case–control and cohort studies were evaluated through National Institute of Health quality assessment tool and Joanna Briggs Institute (JBI) critical appraisal checklist for cohort studies, respectively (Study Quality Assessment Tools and The Joanna Briggs Institute Critical Appraisal tools for use in JBI Systematic Reviews Checklist for Cohort Studies). Assessment of multiple systematic reviews and JBI critical appraisal checklists for RCT were used to assess risk of bias for systematic reviews and RCTs, respectively (The Joanna Briggs Institute Critical Appraisal tools for use in JBI Systematic Reviews Checklist and for Randomized Controlled Trials). 31
Statistical analysis
Assessment of pooled mean and hazard ratio (HR) with 95% confidence interval (CI) was done using binary random-effect model, as appropriate. For continuous variables, Wan et al.’s 32 and Hozo et al.’s 33 statistical methods were used to convert median (IQR) and median (range) to mean ± SD, respectively. We utilized forest plots to illustrate mean PLR levels based on studies reported follow-up, in-hospital, and overall mortality (including follow-up and in-hospital death). We further analyzed PLR mean according to HF status as well as death or survival status. Also, forest plot was depicted to assess the association between PLR and mortality HR. Cochran’s Q statistic, I2, and tau squared (τ2) were measured to evaluate heterogeneity. We evaluated publication bias through funnel plots, Egger’s and Begg’s tests, as well as Duval and Tweedie’s trim-and-fill method. To assess the robustness of the findings, sensitivity analysis using leave-one-out method was done. We used an Excel datasheet to enter all recruited data, and comprehensive meta-analysis software (version 2.0, Biostat, Englewood, NJ) and RStudio software (version 1.1.463, RStudio, Inc., Boston, MA) were statistical tools used to perform the analyses.
Results
We found 482 records during primary literature review. After elimination of 158 duplicated records, 324 records were investigated. Based on inclusion/exclusion criteria, a total of 21 studies on 13,924 patients were selected for final analysis (Figure 1). The total population had a mean age of 70.36 ± 12.88 years (n = 20 studies) and 61.72% were males. Summary of included studies is shown in Table 1. Except for three records, all other articles had cross-sectional designs.34–36 Risk of bias and quality assessments of recruited articles are provided in Supplemental Tables S1 and S2. In total, 20 articles (n = 10,674) reported PLR levels. The mean PLR in total HF population was found to be 165.54 (95% CI: 154.69–176.38) (Figure 2). We also provided results of heterogeneity indices in Table 2. In terms of publication bias, funnel plot is presented in Supplemental Figure S1. Although Begg’s test (p = 0.054) was in favor of no publication bias, the result of Egger’s test was the opposite (p = 0.0004). Duval and Tweedie’s trim-and-fill method revealed potential five missing studies [observed point estimate: 165.54 (95% CI: 154.69–176.38), adjusted point estimate: 151.59 (95% CI: 141.28–161.89)]. To assess the robustness of the findings, a sensitivity analysis was performed, indicating the consistency of the outputs (Supplemental Figure S2).
Summary of included studies reporting PLR and HF clinical outcomes.
CI, confidence interval; CVD, cardiovascular disease; HF, heart failure; IQR, interquartile range; NA, not applicable; NR, not reported; OR, odds ratio; PLR, platelet-to-lymphocyte ratio; Q, quartile; SD, standard deviation; T, tertile.
Forest plot for mean PLR based on total population.
Heterogeneity results of included studies according to PLR.
Cochran’s Q statistic for heterogeneity.
Index for the degree of heterogeneity.
Tau-squared measure of heterogeneity.
HFrEF, heart failure with reduced ejection fraction; PLR, platelet-to-lymphocyte ratio.
PLR and mortality
In total, 18 articles (n = 10,084) were found to report PLR levels and mortality (including either follow-up or in-hospital death) in HF patients.26,27,34–37,39–44,46–51 The mean age was 72.24 ± 12.87 years (males: 61.95%). Figure 3 depicts a forest plot of PLR means in enrolled records reported mortality. The total PLR mean was 166.68 (95% CI: 154.87–178.50). The heterogeneity results are provided in Table 2. The funnel plot is also provided in Supplemental Figure S3. Begg’s (p = 0.046) and Egger’s (p = 0.0006) tests were in favor of potential publication bias. Moreover, the results of Duval and Tweedie’s trim-and-fill method showed the presence of two missing studies [observed point estimate: 166.68 (95% CI: 154.87–178.50), adjusted point estimate: 160.46 (95% CI: 148.98–171.94)]. However, sensitivity analysis confirmed the robustness of our findings (Supplemental Figure S4).
Forest plot for mean PLR based on studies reported mortality (follow-up or in-hospital mortality).
In total, 16 records (n = 8159) reported follow-up mortality in HF sufferers.26,34–37,39–44,46–49,51 Follow-up duration ranged from 30 days to 66 (IQR: 35–105.5) months.35,43 This subset of the population had a mean age of 71.90 ± 13.26 years (males: 62.04%). The mean PLR was found to be 162.55 (95% CI: 149.35–175.75) (Figure 4). Heterogeneity indices were quite considerable (Table 2). For publication bias, the funnel plot is shown in Supplemental Figure S5. Despite Begg’s test did not show any bias (p = 0.196), Egger’s test (p = 0.002) was in favor of potential publication bias. However, the results of Duval and Tweedie’s trim-and-fill method showed similar observed and adjusted point estimates, indicating no missing studies. We also performed the sensitivity analysis to find the impact of any included study on the findings. As shown in Supplemental Figure S6, our findings were robust.
Forest plot for mean PLR based on studies reported follow-up mortality.
Five articles on 2179 subjects reported in-hospital mortality rate (mean age: 73.18 ± 11.16 years, males: 60.34%) with PLR levels of 192.83 (95% CI: 150.06–235.61) (Figure 5).27,40,46,48,50 We provided heterogeneity outputs and the funnel plot in Table 2 and Supplemental Figure S7, respectively. Begg’s and Egger’s p values were 0.403 and 0.035, respectively. In addition, Duval and Tweedie’s trim-and-fill method did not show any missing studies (similar observed and adjusted point estimates). Sensitivity analysis indicated the consistency of the findings (Supplemental Figure S8).
Forest plot for mean PLR based on studies reported in-hospital mortality.
PLR based on survival and death subgroups
We found 13 records (n = 7188) reported PLR means in survived or dead HF patients.26,34,35,37,39,40,42–44,46–49 PLR mean differed significantly between groups in a way that deceased subjects had higher PLR levels in comparison to survived group [194.73 (95% CI: 175.60–213.85) versus 152.34 (95% CI: 134.01–170.68)] (Figure 6). The standard PLR mean difference was also remarkable. Survived patients had significantly lower PLR ranges compared to deceased ones [−0.592 (95% CI: −0.857, −0.326), p < 0.001] (Figure 7). A funnel plot to demonstrate publication bias is depicted in Supplemental Figure S9 with a further report of heterogeneity indices in Table 2. Neither Begg’s (p = 0.292) nor Egger’s (p = 0.169) test was in favor of any publication bias. However, Duval and Tweedie’s trim-and-fill method revealed different observed [−0.592 (95% CI: −0.857, −0.326)] and adjusted [−0.682 (95% CI: −0.941, −0.423)] point estimates, indicating the presence of two missing studies. Complementary sensitivity analysis confirmed the robustness of the outcomes (Supplemental Figure S10).
Forest plot for mean PLR based on studies reported death and survival groups.
Forest plot for standard PLR mean difference in survived versus deceased patients.
PLR and mortality HR
Six articles reported PLR impact (as a continuous variable) on HR of follow-up death.35,38,41,42,44,49 However, two records were not included in the analysis due to inconsistent CIs.35,44 Figure 8 shows a forest plot of PLR effect on mortality. Our findings failed to reveal any significant association between increased PLR and long-term mortality risk (HR: 1.02, 95% CI: 0.99–1.05, p = 0.289) (heterogeneity indices: Table 2).
Forest plot for PLR (as a continuous variable) mortality hazard ratio.
Two studies reported multi-variable adjusted mortality HR based on PLR tertiles.46,48 Data analysis revealed risk of death was not statistically significant across higher tertiles in comparison to the first PLR tertile (T2 versus T1: HR: 1.74, 95% CI: 0.81–3.72, p = 0.156 and T3 versus T1: HR: 2.40, 95% CI: 0.46–12.60, p = 0.300) (Figure 9). Three other studies reported mortality risk according to PLR quartiles.26,34,35 With the exclusion of one study due to inconsistent CIs, 34 complementary analysis did not find any significant association between PLR quartiles (Q2 versus Q1: HR: 0.72, 95% CI: 0.29–1.81, p = 0.483, Q3 versus Q1: HR: 1.01, 95% CI: 0.56–1.83, p = 0.982, and Q4 versus Q1: HR: 1.09, 95% CI: 0.83–1.43, p = 0.523) (Figure 10).
Forest plot of mortality hazard ratio according to PLR tertiles.
Forest plot of mortality hazard ratio according to PLR quartiles.
PLR and rehospitalization
One record reported a readmission rate. Heidarpour et al. implemented a cross-sectional study to investigate the role of PLR in rehospitalization among HF patients. After a follow-up duration of 4.26 ± 2.2 months, 98 out of 405 (24.2%) acute decompensated HF (ADHF) patients were readmitted. The rehospitalization distribution across PLR quartiles were the following: Q1: 24 out of 100 (24%), Q2: 28 out of 103 (27.2%), Q3: 30 out of 101 (29.7%), and Q4: 16 out of 101 (15.8%). However, the difference was not statistically significant (p = 0.111). 26
PLR and HF detection
Two articles evaluated the PLR capability to predict HF. Yurtdaş and colleagues enrolled 40 HF individuals with left ventricular ejection fraction (LVEF) of less than 40% and compared the PLR levels with 30 healthy controls (LVEF ⩾ 50%). Mean PLR differed significantly between patients and controls (179 ± 125 versus 101 ± 82, p = 0.004). Results of multivariate regression analysis suggested PLR could be an HF predictor [odds ratio (OR): 1.015, 95% CI: 1.001–1.028, p = 0.030]. 45 On the other hand, Durmus et al. enrolled 56 decompensated HF regardless of LVEF and 40 controls to assess whether PLR was capable of independently predicting HF. However, multi-variable logistic regression analysis failed to prove any capability in this regard (OR: 0.993, 95% CI: 0.976–1.010, p = 0.407). 51
PLR and HF worsening
One record reported the potential impact of PLR on HF worsening. After a median follow-up of 2.24 (IQR: 1.18–3.97) years, 298 out of 3220 patients experienced worsening of HF, and the multivariate Cox regression model indicated PLR >75th percentile had been associated with 1.50 (95% CI: 1.17–1.93, p = 0.0013) times increased risk of cardiac function deterioration. 28
PLR quartiles and HF outcomes
Four records reported specific PLR quartiles.16,28,34,35 In Wu et al.’s study, the following were defined as PLR quartiles for 1207 HF subjects: Q1: PLR < 89.5, Q2: 89.5 ⩽ PLR < 116.7, Q3: 116.7 ⩽ PLR < 156.9, and Q4: PLR ⩾ 156.9. After the median follow-up for 66 months, they found significantly higher HR of all-cause death among those within the highest quartile compared to the first PLR quartile in the univariate model but not in the multivariate one (1.45, 95% CI: 1.14–1.83, p = 0.002, and 1.09, 95% CI: 0.83–1.44, p = 0.5, respectively). The area under the curve (AUC) for the prediction of all-cause mortality was 0.58 (95% CI: 0.55–0.61) with a sensitivity and specificity of 18.7% and 89.8%, respectively. 35 In a total of 405 ADHF individuals, the following were defined as PLR quartiles: Q1: PLR ⩽ 118 (n = 100), Q2: 119 < PLR < 198 (n = 103), Q3: 198 ⩽ PLR < 268 (n = 101), and Q4: PLR ⩾ 268 (n = 101). They assessed mortality and rehospitalization occurrence during a mean follow-up duration of 4.26 (SD: 2.2) months and 44 (10.9%) deaths and 98 (24.2%) readmissions were observed, respectively [deaths: Q1: 14 (14%), Q2: 8 (7.8%), Q3: 11 (10.9%), and Q4: 11 (10.9%), rehospitalizations: Q1: 24 (24%), Q2: 28 (27.2%), Q3: 30 (29.7%), and Q4: 16 (15.8%)]. However, neither death (p = 0.565) nor readmission (p = 0.111) differed remarkably across PLR quartiles. Multivariate adjusted Cox regression analysis failed to prove any differences between the first PLR quartile compared to the others (Q2 versus Q1: HR: 0.40, 95% CI: 0.16–1.01, p = 0.054, Q3 versus Q1: HR: 0.61, 95% CI: 0.23–1.56, p = 0.305, and Q4 versus Q1: HR: 1.12, 95% CI: 0.34–3.72, p = 0.845). 26 In another study, the followings were considered for PLR quartiles in 443 acute HF individuals [Q1: <110.63 (n = 111), Q2: 110.63–139.23 (n = 112), Q3: 139.23–177.17 (n = 110), and Q4: >177.17 (n = 110)]. The mean follow-up time was 143.68 days (range: 20–180 days) and 160 (36.12%) deaths occurred [Q1: 11 (9.90%), Q2: 26 (23.21%), Q3: 54 (49.09%), and Q4: 69 (62.72%), p < 0.001]. Six-month survival rates according to consecutive PLR quartiles were 90.09%, 76.79%, 50.07%, and 32.27%. Multivariable Cox regression model revealed patients within the third and fourth PLR quartiles had 3.118 (95% CI: 1.668–5.386, p < 0.001) and 2.437 (95% CI: 1.302–3.653, p < 0.001) times higher risk of mortality in comparison to the first quartile, respectively. 34 Another study on 3250 HF subjects indicated that the fourth PLR quartile (>75th percentile) had been associated with 1.50 (95% CI: 1.17–1.93) times increased risk of HF worsening. 28
PLR tertiles and HF outcomes
Four records were found to assess clinical outcomes in HF according to PLR tertiles.27,43,46,48 Delcea and colleagues enrolled 1299 HF patients and categorized them based on PLR tertiles [T1: 14.26–108.02 (n = 433), T2: 108.07–154.78 (n = 433), and T3: 154.80–992.88 (n = 433)]. In total, 37 (2.84%) deaths occurred during hospitalization with the following distribution: T1: 6 (1.4%), T2: 10 (2.3%), and T3: 21 (4.8%), p = 0.006. They also assessed the length of hospital stay and found significant differences between PLR tertiles [T1: median (IQR): 5 (3–7), T2: median (IQR): 5 (4–7), and T3: median (IQR): 6 (4–9)]. 27 Likewise, 115 HF patients who suffered from acute cardiogenic pulmonary edema were selected by Demir and colleagues. PLR tertiles were defined as low [PLR < 98.3 (n = 38)], medium [98.3–194.97 (n = 39)], and high [PLR > 194.97 (n = 38)]. In total, 10 (8.7%) patients died during their admissions [T1: 2 (8%), T2: 3 (11.5%), and T3: 5 (20%)] but the difference was not significant compared to survivors (p = 0.435). They followed participants for 20.8 ± 16.1 months and 39 (37.14%) deaths were observed [T1: 7 (17.94%), T2: 14 (35.89%), and T3: 18 (46.15%)]. Further assessment of total mortality revealed a significant PLR difference between deceased and survived subjects (p = 0.005). Multivariate Cox regression model indicated that patients within second and third PLR tertiles had higher mortality risk compared to the first tertile (T2 versus T1: HR: 2.730, 95% CI: 1.198–6.221, p = 0.017, T3 versus T1: HR: 5.657, 95% CI: 2.467–12.969, p < 0.001). 46 Pourafkari and colleagues determined PLR tertiles as T1: PLR < 137, T2: 137–210, and T3: PLR > 210. During the follow-up, 198 out of 354 HF patients died and Cox regression multivariate analysis for follow-up death did not show any considerable risk differences between the second and third tertiles compared to the first one (T2 versus T1: HR: 1.243, 95% CI: 0.707–2.184 and T3 versus T1: HR: 1.043, 95% CI: 0.497–2.188). 48 Thirty-day mortality rates for 45 out of 439 ADHF patients based on PLR tertiles were the following: T1: 4 (8.9%), T2: 13 (28.9%), and T3: 28 (62.2%). The difference was statistically significant compared to PLR tertiles among survivors (p < 0.001). They found that PLR tertiles could predict 30-day mortality in ADHF patients (multivariate adjusted OR: 1.94, 95% CI: 1.12–3.35, p = 0.018). 43
PLR cutoff and HF outcomes
Six records reported unique PLR cutoff values.27,36,42,43,45,51 Tamaki and colleagues reported a PLR cutoff point of 193 for cardiovascular death prediction in 1026 HF with preserved ejection fraction (HFpEF) patients (AUC: 0.60, 95% CI: 0.57–0.63, p = 0.0009, sensitivity: 56%, specificity: 62%). They found that patients who had higher PLR values experienced cardiovascular and all-cause death more frequently (HR: 1.76, 95% CI: 1.09–2.84, p = 0.0215, and HR: 1.47, 95% CI: 1.06–2.03, p = 0.0198, respectively). 36 Delcea et al. 27 found PLR cutoff point of 154.78 could reliably predict in-hospital mortality (sensitivity: 54.05%, specificity: 67.40%, AUC: 0.658, 95% CI: 0.567–0.750, p = 0.001) and extended length of hospital stay, defined as hospitalization for at least 7 days (sensitivity: 49.13%, specificity: 71.43%, AUC: 0.626, 95% CI: 0.587–0.666, p < 0.001). The cutoff was reported to be 150 in Sadeghi et al.’s study (sensitivity: 51.7%, specificity: 72.3%, positive predictive value: 25.9%, negative predictive value: 88.9%, accuracy: 69%, AUC: 0.62, p = 0.041). In multivariate Cox regression analysis, PLR did not show any predictability for 30-day survival (HR: 1.006, 95% CI: 0.99–1.021, p = 0.487). 42 In another study, 272.9 was set to be the optimal PLR cutoff (AUC: 0.71) and patients within the higher PLR group (n = 94) had 3.22 (95% CI: 1.56–5.68, p < 0.001) times increased odds of 30-day mortality compared to the lower group (n = 345). Also, increased PLR was mostly observed in deceased subjects compared to the survivors (55.6% versus 17.5%, p < 0.001). 43 Yurtdaş et al. 45 found the PLR cutoff point of 73 (sensitivity: 83%, specificity: 53%, AUC: 0.76, 95% CI: 0.65–0.88, p < 0.001) as an optimal value to predict HF. Another HF predicting cutoff point was found to be 137.3 in Durmus et al.’s study (sensitivity: 70%, specificity: 60%, AUC: 0.689, p = 0.004). 51
PLR and HF status
Seven studies (n = 2151) investigated the PLR impact on HF patients with reduced ejection fraction (HFrEF).37,40–42,44,45,47 In terms of HFpEF, there was only one study 36 and others included all HF patients regardless of LVEF.26–28,34,35,37–39,43,46,48,49,51 Figure 11 presents the forest plot for PLR mean in HFrEF individuals. The mean was 144.96 (95% CI: 130.38–159.54). Heterogeneity indices are provided in Table 2. Neither Begg’s nor Egger’s test results indicated publication bias [p = 0.274, and p = 0.077, respectively, funnel plot (Supplemental Figure S11)]. However, Duval and Tweedie’s trim-and-fill method suggested that two studies were missing (observed point estimate: 144.96, 95% CI: 130.38–159.54, adjusted point estimate: 134.48, 95% CI: 119.88–149.08). To assess the impact of each study on the findings, we performed a sensitivity analysis and the results supported the robustness of our results (Supplemental Figure S12).
Forest plot for mean PLR in patients with heart failure with reduced ejection fraction.
Five studies reported PLR values among survived and deceased subjects who suffered from HFrEF. Complementary analyses revealed live patients had significantly lower PLR mean compared to the death group (standard mean difference: −0.979, 95% CI: −1.384 to −0.573, p < 0.001) (Figure 12). Table 2 shows the heterogeneity indices of the included studies. The funnel plot, as shown in Supplemental Figure S13, did not reveal any publication bias, also confirmed by Begg’s (p = 0.110) and Egger’s (p = 0.152) tests. Duval and Tweedie’s trim-and-fill method was in favor of one missing study (observed point estimate: −0.979, 95% CI: −1.384 to −0.573, adjusted point estimate: −1.090, 95% CI: −1.461 to −0.718). Sensitivity analysis indicated our findings were robust and consistent (Supplemental Figure S14).
Forest plot for standard PLR mean difference in survived versus deceased heart failure with reduced ejection fraction patients.
Discussion
The main aim of this systematic review and meta-analysis was to evaluate the impact of PLR on clinical outcomes among patients suffering from HF. The PLR mean in the total HF population was determined to be 165.54 (95% CI: 154.69–176.38). We also found patients who died in the context of HF had significantly higher PLR levels rather than survivors [194.73 (95% CI: 175.60–213.85) versus 152.34 (95% CI: 134.01–170.68)]. This trend was also observed for patients with HfrEF. However, further analysis was not significant in terms of PLR association with mortality risk, either as a continuous or as a categorical (tertiles or quartiles) variable. Since HF prevalence increases as people age and it is associated with considerable mortality rate and poor prognosis despite HF management progress, this index should be used cautiously to evaluate clinical outcomes and subsequently prioritize high-risk individuals.
Inflammation has been reported to play pivotal roles in the pathogenesis of CVDs through different mechanisms including oxidative stress and cell-mediated immune responses.52,53 Although the exact pathophysiological mechanism for HF is still not defined, several theoretical mechanisms are proposed. The first one is related to inflammatory cytokines. Inflammation leads to the production and secretion of various cytokines like interleukin (IL)-4, IL-6, C-reactive protein, and tumor necrosis factor-α.54–56 These cytokines result in negative effects on cardiac cells ultimately leading to decreased cardiac pump function.54,55 Also, different cells are involved in HF pathogenesis. Some culprit cells are platelets, monocytes, and neutrophils. They initiate inflammatory cytokine production and secretion, involve other cell lines, and alter receptor expression to induce additional inflammatory markers to continue disease progression.57,58 Platelets can interact with both leukocytes and endothelial cells. They perform their functions by production of inflammatory factors, subsequently leading to monocyte transmigration. They are called to be a link between thrombosis, inflammation, and atherosclerosis pathogenesis.59,60 In addition to increased platelet counts due to inflammatory cytokines, thrombocytopenia was also observed in HF patients. Getawa and Bayleyegn’s 61 study on 245 HF sufferers showed thrombocytopenia was a more prevalent platelet abnormality in comparison to thrombocytosis [12.24% (95% CI: 8.67–17.01%) versus 2.86% (95% CI: 1.36–5.90%)]. Lower platelets have been reported to be an independent prognostic tool. Mojadidi et al. 62 found HF individuals with moderate to severe thrombocytopenia (platelets <100,000/µl) had 1.84 (95% CI: 1.33–2.56, p < 0.001) times higher 1-year all-cause mortality risk in comparison to those with normal or mild thrombocytopenia. Although the exact etiology of this phenomenon remains to be elucidated, one possible explanation might be related to abnormal immune response due to medication therapy. 63 It seems both increased or decreased platelet counts might be effective in HF prognosis. However, data are still limited, and further studies are warranted.
On the other hand, lymphocytes have been reported to be a protective factor in inflammation through tissue inhibitors of metalloproteinase-1 expression. 64 However, lymphocyte counts could be reduced during physiologic stress. High cortisol and catecholamine secretion through the hypothalamus–hypophysis–adrenal axis during a stressful condition, like HF, could reduce lymphocytes by redistribution of them to lymphatic organs, finally leading to a decrease in survival rate.26,65 Moreover, HF sufferers have lower flow in coronary vessels and higher platelets could lead to a pro-thrombotic state and worsen the prognosis.37,66 Therefore, high platelet counts and low lymphocytes result in raised PLR and this inexpensive blood index might be practical in clinical settings to assess prognosis.
Another considerable factor is HF subtypes. We found a PLR mean of 144.96 (95% CI: 130.38–159.54) in HFrEF and deceased subjects had higher values in comparison to the survived subjects. Despite the probable role of inflammation in HF has been widely investigated in recent years, the exact mechanism might differ according to HF types. For instance, one of the main triggers in HFrEF is myocardial injury, resulting in the initiation of inflammatory responses mediated by pro-inflammatory cytokines as well as the recruitment of immune cells. Subsequent collagen synthesis by myofibroblasts causes scar formation and reduction in LVEF.67,68 This inflammatory response might be continued because of sustained myocardial injury or other mechanisms irrespective of tissue injury, including hemodynamic overload, activation of the renin–angiotensin–aldosterone, or sympathetic nervous system. 68
In terms of HFpEF, some underlying comorbidities, such as diabetes mellitus, hypertension, obesity, chronic kidney diseases, and chronic obstructive pulmonary disease initiate an inflammatory state leading to decreased nitric oxide bioavailability and increased production of reactive oxygen species. Further infiltration of blood monocytes into the cardiac tissue and their differentiation into macrophages cause impaired ventricular relaxation.69,70
The role of biological pathways in different HF types is still being investigated. Tromp et al. evaluated 92 unique biomarkers in 1544 HF patients to find the correlations in HFrEF, HFpEF, and HF with midrange ejection fraction (HFmrEF). Network analysis revealed HFpEF biomarkers were mostly associated with inflammatory as well as extracellular matrix reorganization pathways. By contrast, common pathways in HFrEF were cardiac stretch, metabolism, and cellular proliferation. In terms of HFmrEF, the pathways were in between HFrEF and HFpEF.71,72 Some specific inflammatory-related biomarkers suggested for HFpEF include Pentaxin-3 and receptor for the enhanced glycation end product. 72 These result in interstitial fibrosis and cardiac cells stiffening, subsequently causing increased left ventricular filling pressure.69,73
PLR has been suggested to negatively affect clinical outcomes in CVDs. A systematic review and meta-analysis indicated higher PLR had been associated with higher risk of both in-hospital and long-term major cardiovascular events in patients with acute coronary syndrome (ACS) [relative risk (RR): 1.95, 95% CI: 1.30–2.91, p = 0.001 and RR: 1.50, 95% CI: 1.08–2.09, p = 0.01, respectively]. 74 This index has also been associated with the severity of atherosclerosis in ACS sufferers. 75
In the current study, we found higher PLR was mostly observed in deceased HF patients rather than survivors with a significant standard mean difference in alive versus dead patients (−0.592 (95% CI: −0.857 to −0.326), p < 0.001], and assessment of this index might be helpful for prognosis determination in clinical wards. However, complementary studies are still required. In terms of hospitalization, only one record reported no potential PLR effect on readmission among HF patients. 26 Since readmission poses a considerable economic burden, implementation of complementary studies is warranted to clarify this association. For HF detection, two records reported opposite outcomes.45,51 In one study, PLR was found to be an acceptable HF predictor in comparison to healthy controls (OR: 1.015, 95% CI: 1.001–1.028). 45 On the other hand, PLR was not able to reliably predict HF in another record (OR: 0.992, 95% CI: 0.976–1.010, p = 0.407). 51 Although the different HF types might be a potential explanation for this inconsistency, further studies are required to exactly evaluate the predictability of PLR for HF diagnosis. PLR and worsening of HF were evaluated in one record and the outcomes were in favor of increasing cardiac function deterioration with PLR > 75th percentile. However, no significant difference was found between two HF subtypes, including HFrEF and HFpEF. 28 Although one record is insufficient to generalize the outcomes, further studies might help to establish the PLR effect on HF worsening in different HF phenotypes.
PLR has been reported in different ways to assess HF clinical outcomes in multiple records including quartiles, tertiles, and optimum cutoff points. Different quartiles were the followings in recruited records: Q1: PLR < 89.5, Q2: 89.5 ⩽ PLR < 116.7, Q3: 116.7 ⩽ PLR < 156.9, and Q4: PLR ⩾ 156.9, Q1: PLR ⩽ 118, Q2: 119 < PLR < 198, Q3: 198 ⩽ PLR < 268, Q4: PLR ⩾ 268, and Q1: <110.63, Q2: 110.63–139.23, Q3: 139.23–177.17, Q4: >177.17.26,34,35 PLR tertiles were also defined in three studies (T1: 14.26–108.02, T2: 108.07–154.78, T3: 154.80–992.88 and T1: PLR < 98.3, T2: 98.3–194.97, T3: PLR > 194.97 and T1: PLR < 137, T2: 137–210, T3: PLR > 210).27,46,48 Moreover, different PLR cutoff points of 193, 154.78, 150, 272.9, 73, and 137.3 with different ranges of sensitivity, specificity, and AUC values have been reported to be used in HF detection and outcome assessment.27,36,42,43,45,51 This quite wide PLR cutoff range might be associated with different sample sizes and necessitates complementary studies to define the exact cutoff point.
This systematic review and meta-analysis is the first in the literature to evaluate the potential impact of PLR on HF outcomes. We tried our best to include every relevant record with no time limitation to increase the generalizability of our findings. However, some limitations are still present. We recruited only English-published articles and did not screen other non-English records and we were unable to compare PLR means based on gender. Enrolled studies had different sample sizes and study designs which disabled us from performing statistical analysis with low heterogeneity. Further analysis according to HF severity was not feasible due to the low number of records in each category. We were able to perform subgroup analysis on HFrEF patients and PLR comparison with other HF types was not feasible. Although no significant association was found in terms of PLR and mortality risk, the few number of recruited articles as well as the definition of different PLR tertiles or quartiles might explain this phenomenon.
In conclusion, despite PLR being a simple widely available tool and might be a good choice in nations with limited healthcare resources, this ratio should be investigated further for its predictability of HF clinical outcomes, and several other studies are warranted to explore this association, especially in terms of mortality risk.
Supplemental Material
sj-doc-2-tak-10.1177_17539447241227287 – Supplemental material for The impact of platelet-to-lymphocyte ratio on clinical outcomes in heart failure: a systematic review and meta-analysis
Supplemental material, sj-doc-2-tak-10.1177_17539447241227287 for The impact of platelet-to-lymphocyte ratio on clinical outcomes in heart failure: a systematic review and meta-analysis by Mehrbod Vakhshoori, Niloofar Bondariyan, Sadeq Sabouhi, Keivan Kiani, Nazanin Alaei Faradonbeh, Sayed Ali Emami, Mehrnaz Shakarami, Farbod Khanizadeh, Shahin Sanaei, Niloofaralsadat Motamedi and Davood Shafie in Therapeutic Advances in Cardiovascular Disease
Supplemental Material
sj-docx-1-tak-10.1177_17539447241227287 – Supplemental material for The impact of platelet-to-lymphocyte ratio on clinical outcomes in heart failure: a systematic review and meta-analysis
Supplemental material, sj-docx-1-tak-10.1177_17539447241227287 for The impact of platelet-to-lymphocyte ratio on clinical outcomes in heart failure: a systematic review and meta-analysis by Mehrbod Vakhshoori, Niloofar Bondariyan, Sadeq Sabouhi, Keivan Kiani, Nazanin Alaei Faradonbeh, Sayed Ali Emami, Mehrnaz Shakarami, Farbod Khanizadeh, Shahin Sanaei, Niloofaralsadat Motamedi and Davood Shafie in Therapeutic Advances in Cardiovascular Disease
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
