FormalPara Key Summary Points

Thrombocytopenia has been reported with both linezolid and vancomycin and can lead to treatment interruption

In this retrospective study, linezolid was associated with an approximately four times higher incidence of thrombocytopenia compared to vancomycin

Risk factors for development of thrombocytopenia included advanced age, low baseline platelet count and concomitant use of vasopressors

There is a variation in frequency of linezolid-associated thrombocytopenia in our study compared to the studies from other regions such as the USA and Canada; future studies focusing on establishing pharmacogenomic links are suggested

Introduction

Linezolid is the first antibacterial drug in the class of oxazolidinones. It inhibits the activity of a broad range of gram-positive bacteria, including methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE), glycopeptide-intermediate S. aureus (GISA) and penicillin-resistant Streptococcus pneumoniae [1]. Because it is available in an oral dosage form, linezolid is typically preferred for patients healthy enough to be treated on an outpatient basis. However, its adverse effects include peripheral neuropathy, increased liver enzymes, serotonin syndrome and myelosuppression [2]. Considering the global threat of antimicrobial resistance, linezolid has an important place among therapeutic antimicrobial options so that gaining further insights into its toxicities and identifying at-risk patients are crucial for optimizing therapy [2]. Vancomycin’s spectrum closely resembles that of linezolid. This tricyclic glycopeptide antibiotic, which is used to treat and prevent many bacterial infections caused by gram-positive bacteria, including MRSA, is also used to treat streptococci, enterococci and methicillin-susceptible Staphylococcus aureus (MSSA) infections [3]. Linezolid and vancomycin are important treatments of multidrug-resistant infections in hospitalized patients, and thrombocytopenia has been reported with both [1, 3].

Thrombocytopenia, or a platelet count < 150,000/µl, can be triggered by many factors, including cancer, aplastic anemia, alcohol, toxic chemicals, medications, viruses and genetic conditions [4]. It can lead to life-threatening bleeding, petechiae, purpura or mucosal bleeding (e.g., epistaxis) [4]. In the intensive care unit (ICU), it can restrict patient management and may call for withholding or delaying necessary invasive interventions, decreasing anticoagulation dose, preparing platelet transfusion and/or changing anticoagulants [5]. The most commonly implicated drugs used in the ICU are heparin and antibiotics [5].

In the phase III approval trial for linezolid, the frequency of thrombocytopenia was reported to be 2.4% with a treatment duration of 28 days [2]. Soon after approval, reports surfaced of a higher incidence in clinical practice, reaching 48% [6,7,8,9,10]. A significant number of studies reporting such a higher incidence involved Japanese patients, but whether this phenomenon reflects a higher rate of observation in this population or simply reflects research interest remains unclear [7,8,9, 11,12,13]. Despite the wide use of vancomycin, the incidence of vancomycin-induced thrombocytopenia remains unknown. However, based on case reports and two retrospective studies, a prevalence of 5.9–7.1% was reported, which was higher than expected. Nevertheless, the variation in thrombocytopenia definitions among these studies and their retrospective design might have influenced results [14,15,16].

Nasraway et al. attempted to compare the incidence of thrombocytopenia between vancomycin and linezolid by analyzing data from two separate clinical trials based on similarities in their population and design; they reported that linezolid was not associated with thrombocytopenia more frequently than with vancomycin. Considering the low incidence of such adverse effects in a clinical trial setting, the pooled analysis interpretation has its limitations [17].

Besides the reported incidences of thrombocytopenia from controlled trials, no studies have compared the incidence of linezolid and vancomycin-induced thrombocytopenia in a real-world setting. The primary objective of this study was to compare the incidence of thrombocytopenia in the linezolid and vancomycin groups. The secondary objective was to evaluate the risk factors associated with development of thrombocytopenia.

Methods

This retrospective observational cohort study involved patients who received linezolid (intravenously or orally) or vancomycin (intravenously) at the Ministry of National Guard Hospital in Riyadh, Saudi Arabia, between January 2016 and October 2019 for a minimum of 5 days and whose platelet values were measured during treatment.

Patients were identified from the hospital’s electronic health care database, and data collected included demographics, comorbidities, comorbidity index (using Charlson comorbidity index), concomitant drugs, indication for either vancomycin or linezolid, microbiological data, laboratory data and disease severity.

The laboratory values collected included baseline values (closest to initiation of linezolid or vancomycin) and the most divergent values seen during treatment.

Thrombocytopenia was defined as a platelet count < 150,000/µl and further subdivided based on severity into platelet counts < 150,000/µl, < 100,000/µl and < 50,000/µl.

Patients known to have hematological malignancy such as leukemia or myelodysplastic syndrome, with confirmed heparin-induced thrombocytopenia, who received chemotherapy and with a baseline abnormal platelet count < 150,000/µl prior to therapy were excluded.

The sample size was estimated using the previous literature [16,17,18]. The incidence rate of thrombocytopeina in the linezolid and vancomycin treatment groups was 37.7% and 22.3%, respectively, with a 0.05 alpha level and a power of 90%. A total of 450 patients were required, with a minimum of 200 patients in each group.

Statistical analysis was performed using commercially available software SPSS 21.0 and R: A language and environment for statistical computing. Baseline characteristics between patients receiving linezolid and patients receiving vancomycin were compared. The association between each continuous variable and thrombocytopenia was tested using Student's t-test. The association between each categorical variable and thrombocytopenia was tested using chi-square test or Fisher's exact test. p < 0.05 was considered to indicate a statistically significant difference. The significant independent variables were included in the final multivariate logistic regression model to compare the association between these significant variables and the development of thrombocytopenia. A sensitivity analysis was added by applying a bootstrapped population from the original sample with the different thrombocytopenia severity levels [19].

The study protocol was reviewed and approved by the Institutional Research Ethics Board (#RC19/254/R). Individual patient consent was not required for a study of this type.

Results

Descriptive Analysis

Of 1420 patients identified in the database, 453 met the inclusion criteria, 241 received linezolid and 212 received vancomycin for ≥ 5 days. The baseline characteristics of patients in both groups are presented in Table 1. Patients in the linezolid group were slightly older and had a higher Charlson comorbidity index, but weight, baseline platelet count and renal function were similar between the two groups.

Table 1 Baseline characteristics

The overall incidence of thrombocytopenia was 20.1% in our study group. Of these, 30.3% were in the linezolid group and only 8.5% of patients in the vancomycin group (p < 0.001). Of the patients in the linezolid group that developed thrombocytopenia, 30.3% had a platelet count < 150,000/µl, 13.7% had a platelet count < 100,000/µl and 4.6% had a platelet count < 50,000/µl. In the vancomycin group the platelet count was  < 150,000/µl in 8.5%, < 100,000/µl in 3.78% and < 50,000/µl in 1.89% of the patients. The significant independent variables associated with thrombocytopenia are presented in Tables 2 and 3.

Table 2 Association between continues variables and thrombocytopenia
Table 3 Association between categorical variables and thrombocytopenia

The main logistic regression analysis revealed that patients in the linezolid group had an approximately four times higher risk of thrombocytopenia (OR 4.39; 95% CI 2.38–8.08) and had a 3.8 times higher risk of having a platelet count < 100,000/µl (OR 3.77; 95% CI 1.55–9.17) compared to the vancomycin group (Table 4). In a bootstrapped sample, linezolid therapy was associated with a statistically significant 75% (OR 1.75; 95% CI 1.68–1.83) elevated risk of severe thrombocytopenia (platelets < 50,000/µl) compared to vancomycin group. Table 5 shows original and bootstrapped samples for different stages of thrombocytopenia.

Table 4 Logistic regression for the association between significant variables and the risk of thrombocytopenia
Table 5 Sensitivity analysis with different values for thrombocytopenia with different population

Risk Factors

Patients with thrombocytopenia had a statistically significant older age, higher Charlson comorbidity index score and lower baseline platelet count (see Table 2). After controlling for other factors, only age and baseline platelet count were associated significantly with thrombocytopenia (OR 1.02; 95% CI 1.02–1.03; p < 0.03) and (OR 0.99; 95% CI 0.93–0.98; p < 0.0002), respectively (Table 4).

In the logistic regression, only vasopressor therapy showed an approximately 2.5 times higher incidence of thrombocytopenia (OR 2.39; 95% CI 1.23–4.66; p 0.01). No significant association was seen with piperacillin-tazobactam, penicillin, cephalosporin, trimethoprim sulfamethoxazole, ranitidine, GPIIb-IIIa Inhibitors, non-steroidal anti-inflammatory drugs (NSAIDs), carbamazepine and phenytoin (p > 0.05 for all variables). The non-significant variables are presented in Appendix 1.

Onset of Thrombocytopenia

The average duration of treatment to onset of severe thrombocytopenia in the linezolid group was 14 ± 9.3 days compared with 8.5 ± 6.9 days in the vancomycin group.

Discussion

Multiple studies have used vancomycin as a reference to compare the incidence of linezolid-induced thrombocytopenia. The first by Nasraway et al. was a pooled analysis that compared the incidence of thrombocytopenia reported in two randomized clinical trials (RCTs) of linezolid and vancomycin in patients with nosocomial pneumonia; they described no significant difference in thrombocytopenia, but their study had multiple limitations, and the baseline platelet value for patients on linezolid for > 14 days was only available for 7% of patients [17].

Several preceding RCTs [20, 21] and meta-analyses of RCTs [22, 23] that randomized patients to linezolid or vancomycin reported a higher incidence of thrombocytopenia. The most recent meta-analysis [23] reported a significant difference in the incidence of thrombocytopenia (RR = 20.4, 95% CI 3.12–133.57), while data from observational studies reported conflicting results, reporting no significant difference [24,25,26] in a USA population and those that reported a significantly higher incidence with linezolid vs. vancomycin mainly in an Asian population [16, 27].

In a recent study from Canada, a modest incidence of 17.6% was reported in a retrospective review of 102 inpatients at a tertiary care hospital [28], and in another from France on patients admitted in the ICU only 13.8% developed thrombocytopenia; these lower rates could be due to the stricter definition of thrombocytopenia used in their study [29]. However, as in some inherited genetic disorders that can cause immune thrombocytopenia, a genetic predisposition to this drug-induced adverse effect can be expected; in fact, the conflicting results may allude to a genetic predisposition to this adverse effect [30]. Notably, a recent study reported a significant effect of abcb1 c.3435C>T polymorphism on linezolid clearance [31, 32].

Our results are consistent with most of the studies, showing a significantly higher incidence of thrombocytopenia in association with linezolid compared to vancomycin. In addition, the overall frequency was relatively similar to that reported in the Asian population. Conservatively, when only considering moderate to severe levels of linezolid-associated thrombocytopenia, our results may be closer to a recent study in France by Cazavet et al. [29].

The exact mechanism of linezolid-induced thrombocytopenia has not yet been elucidated. Proposed mechanisms include suppression of release of platelets from mature megakaryocytes [33], oxidative damage to circulating platelets [34] and immune-mediated platelet destruction [35, 36].

Risk factors reported are renal insufficiency, hemodialysis, prolonged treatment duration, elevated or low baseline leukocyte concentration, chronic liver disease and low baseline serum protein concentration [1, 5, 9].

There is also a notion that the adverse effect is dose dependent, as a higher trough concentration of linezolid in the plasma and higher values for area under the concentration-time curve (AUC) have been suggested as possible risk factors [11, 12].

Nevertheless, the pharmacokinetics of linezolid have been reported to be more influenced by a patient’s renal function than their weight [37], with some studies recommending higher doses of linezolid in obese patients. In our study, the mean BMI was 27 in patients that developed thrombocytopenia, and we only included patients that were on the same dose [37, 38].

Concerning renal impairment, only 20% of the patients that developed linezolid-associated thrombocytopenia had an eGFR < 90, and renal impairment was not a significant risk factor. Therefore, the idea of a higher drug exposure might not fully explain the higher frequency seen in our study, but further studies on this subgroup of patients are needed.

An immune-mediated platelet destruction by linezolid has been reported; in these cases, linezolid-related antibodies were detected, and response to immune globulin therapy was observed [35, 39]. The reports also describe a rapid and severe drop in platelets, which was not directly observed in our study where the average duration of treatment to onset of severe thrombocytopenia was 14 ± 9.3 days.

The only factors in our study that significantly increased the risk of thrombocytopenia were the presence of comorbidities and concomitant intake of drugs commonly used in critical patients (e.g., vasopressors). Although vasopressor use might have served as a confounder to the higher rates of thrombocytopenia in the linezolid group, the difference was small at 31% vs. 24% for linezolid and vancomycin, respectively, and cannot be fully attributed to the higher rate of thrombocytopenia in the linezolid group. Therefore, we recommend more frequent monitoring in patients with comorbid conditions that increase their risk of thrombocytopenia.

The weak association of risk factors to predict this adverse reaction and the wide use of linezolid are hurdles in implementing therapeutic drug monitoring (TDM). Considering a variation in frequency of linezolid-associated thrombocytopenia in our study compared to results from the USA and Canada, we suggest that future studies focus on establishing pharmacogenomics links, and healthcare systems should implement good pharmacovigilance practices for monitoring such adverse events [40].

Our study is limited by its retrospective and observational design; in addition, the doses of linezolid and vancomycin were not utilized to identify the role of a dose response and its effect on the incidence of thrombocytopenia in each group. We hypothesized that this adverse reaction is a type B reaction, likely immune mediated and host dependent, based on previous reports. Further studies should stratify patients by onset of thrombocytopenia and expected mechanism of this adverse effect.

Conclusion

Clinicians considering vancomycin or linezolid for a susceptible infection should weigh the higher risk of thrombocytopenia that may be observed with linezolid vs. vancomycin in their decision.

Appendix 1

List of independent variables that were non-significantly associated with the risk of thrombocytopenia

Factor

Thrombocytopenia

P value

No

Yes

Body mass index

29.46

27.32

0.34

Length of hospital stay

65.00

76.41

0.43

Duration of antibiotic

11.1326

12.8571

0.07

Gender

0.56

Female

139

38

 

Male

223

53

 

Piperacillin tazobactam

0.63

No

213

51

 

Yes

149

40

 

Penicillin

0.75*

No

348

89

 

Yes

14

2

 

Cephalosporin

0.37

No

290

69

 

Yes

72

22

 

Trimethoprim sulfamethoxazole

0.9*

No

356

90

 

Yes

6

1

 

Ranitidine

0.43

No

338

87

 

Yes

24

4

 

GPllb–llla inhibitors

0.66*

No

356

89

 

Yes

6

2

 

NSAID

0.27

No

245

67

 

Yes

117

24

 

Carbamazepine

0.9*

No

358

90

 

Yes

4

1

 

Phenytoin

0.79*

No

344

86

 

Yes

18

5

 

Pneumonia

0.25

No

242

55

 

Yes

120

36

 

UTI

0.10

No

294

67

 

Yes

68

24

 

Osteomyelitis

0.59

No

329

81

 

Yes

33

10

 

Intra-abdominal infections

0.15*

No

358

88

 

Yes

4

3

 

Cellulitis

0.39*

No

329

86

 

Yes

33

5

 

Bacteremia

0.29

No

313

83

 

Yes

49

8

 

Line-related sepsis

0.75*

No

348

89

 

Yes

14

2

 

Endocarditis

0.27*

No

354

87

 

Yes

8

4

 

Streptococcus sp.

0.9*

No

351

89

 

Yes

11

2

 

Staphylococcus sp.

0.89

No

216

55

 

Yes

146

36

 

Enterococcus sp.

0.68

No

320

79

 

Yes

42

12

 
  1. *Fisher exact test