Successful Management of Infective Endocarditis Utilizing Rapid High-performance Liquid Chromatography-based Therapeutic Drug Monitoring: Balancing Linezolid Efficacy and Thrombocytopenia Risk

Abstract

Linezolid (LZD) is an effective alternative to vancomycin (VCM) for the treatment of methicillin-resistant Staphylococcus-associated infective endocarditis (IE), including cases involving prosthetic materials. However, the prolonged use often leads to thrombocytopenia, thus highlighting the importance of therapeutic drug monitoring (TDM) for dose optimization. We herein present the case of slowly progressive methicillin-resistant Staphylococcus epidermidis IE affecting a mitral annuloplasty ring in a 69-year-old afebrile patient whose only manifestation was a recurrent cerebral embolism. TDM-guided LZD dose adjustments using rapid high-performance liquid chromatography enabled the successful completion of a six-week antibiotic course while preventing severe thrombocytopenia and ensuring no further stroke recurrence.

Introduction

Linezolid (LZD) is an alternative therapy for methicillin-resistant Staphylococcus-associated infective endocarditis (IE) involving prosthetic valves if vancomycin (VCM) is contraindicated or not tolerated because of adverse events including allergic reactions (1,2). However, thrombocytopenia due to myelosuppression is common after long-term drug administration (3). We herein report a case of slowly progressive IE caused by methicillin-resistant Staphylococcus epidermidis (MRSE) that affected a mitral annuloplasty ring with recurrent cerebral embolism. The infection was successfully managed with LZD dose adjustments guided by therapeutic drug monitoring (TDM), based on plasma concentrations measured using rapid high-performance liquid chromatography (HPLC).

Case Report

The patient was a 69-year-old man who had undergone mitral valvuloplasty five years earlier. He developed an initial cerebral infarction approximately 18 months before referral to our hospital and also experienced multiple recurrent embolic strokes in the following months. Direct oral anticoagulant (DOAC) therapy was initiated at another hospital based on a presumed diagnosis of paroxysmal atrial fibrillation as the embolic source. However, he remained asymptomatic with no palpitations, and no episodes of atrial fibrillation were detected on electrocardiographic monitoring. The patient remained afebrile, and laboratory tests revealed no signs of systemic inflammation. As the source of the embolism remained unidentified, the patient was referred to our hospital for further evaluation.

On admission, his height was 165 cm, weight was 58 kg, temperature was 36.2°C, blood pressure was 108/70 mmHg, heart rate was 64 beats/min, respiratory rate was 12 breaths/min, and SpO₂ was 97% on ambient air. A neurological examination revealed no sequelae of either cerebral infarction or gross paralysis. No petechial hemorrhages, Janeway lesions, osler nodes, or subungual hemorrhages were observed. A cardiac examination revealed a Levine III/VI pansystolic murmur at the apex, with no peripheral edema. Laboratory findings included a white blood cell count of 4,900 /μL, a hemoglobin of 10.9 g/dL, and a platelet count of 156,000 /μL. D-dimer was 0.6 μg/mL, showing no elevation, and the C-reactive protein level was only slightly elevated at 0.6 mg/dL. The liver and kidney functions were normal, and there was no evidence of diabetes. The levels of tumor markers, including CA19-9, carcinoembryonic antigen (CEA), and prostate-specific antigen (PSA), were normal. Autoimmune markers showed negative antinuclear antibody test results and myeloperoxidase-antineutrophil cytoplasmic antibody (MPO-ANCA), ＜1.0 IU/mL, while proteinase 3 (PR3)-ANCA was elevated at ＞350 IU/mL.

Chest radiography revealed a cardiothoracic ratio of 46% with no cardiac enlargement or pulmonary congestion. Electrocardiography revealed a heart rate of 64 bpm with an ectopic atrial rhythm (Fig. 1). Brain magnetic resonance imaging upon admission revealed multiple scattered bilateral cortical infarcts (Fig. 2), with no evidence of mycotic aneurysm or hemorrhaging. Transthoracic echocardiography initially revealed a small, cord-like mobile structure near the mitral annulus; however, it was difficult to differentiate vegetation from normal anatomical structures such as chordae tendineae. Therefore, transesophageal echocardiography (TEE) was performed upon admission, which revealed moderate mitral regurgitation (MR) and mobile vegetation measuring 5 mm and 16 mm on the prosthetic mitral valve ring (Fig. 3). The degree of MR did not worsen compared to the examination performed one year earlier at the referring hospital.

Figure 1.

(a) A chest radiograph and (b) electrocardiogram upon admission. (a) The cardiothoracic ratio is 46% with no cardiac enlargement or pulmonary congestion. (b) The heart rate is 64 beats/min with atrial rhythm.

Figure 2.

Head diffusion-weighted magnetic resonance imaging upon admission. Multiple scattered bilateral cortical infarcts (arrows) are observed.

Figure 3.

Transesophageal echocardiography of the mitral valve upon admission. (a) Two-dimensional 60° view and (b) three-dimensional en-face view of the mitral valve after valvuloplasty. Cord-like vegetations measuring 5 mm and 16 mm (arrows) are visible on the prosthetic mitral valve ring. Moderate mitral regurgitation is observed, with the regurgitant jet colliding with the prosthetic ring at the site of vegetation attachment. LA: left atrium, LV: left ventricle

Three sets of blood cultures obtained on day 1 of hospitalization grew gram-positive cocci; however, delayed microbial growth hindered early identification. Staphylococci grown in blood cultures were identified as Staphylococcus epidermidis, with 99.9% confidence using mass spectrometry analysis with a VITEK MS Prime system (bioMérieux, Marcy l'Etoile, France). As the isolate exhibited poor growth, antimicrobial susceptibility testing using the broth microdilution method was inconclusive. A genetic analysis using the GeneXpert system (Beckman Coulter, Brea, United States) detected the presence of mecA, leading to a diagnosis of MRSE. The patient had poor oral hygiene and had not received regular dental care prior to admission. A dental examination revealed multiple caries and severe periodontitis. These findings are considered potential entry portals for MRSE. The skin was chapped and atopic-like, but it showed no signs of active infection. The patient had no history of steroid or immunosuppressive drug use.

Clinical course

Nonbacterial thrombotic endocarditis (NBTE) associated with malignancy or autoimmune disease was initially considered as a differential diagnosis. The DOAC were discontinued and replaced with continuous intravenous heparin infusion for suspected NBTE. However, D-dimer levels were not elevated and malignancy was ruled out using thoracoabdominal contrast-enhanced computed tomography, esophagogastroduodenoscopy, and colonoscopy. Additionally, PR3-ANCA was positive; however, there were no clinical findings suggestive of vasculitis. After several days of heparin infusion, echocardiography revealed no reduction in vegetation size; therefore, NBTE was impossible.

Despite being afebrile, the patient met the 2023 Duke International Society for Cardiovascular Infectious Diseases Criteria for IE, based on positive blood cultures and TEE findings. The most likely diagnosis was subacute IE involving a prosthetic annuloplasty ring caused by MRSE. The patient was started on triple therapy with VCM, gentamicin (GM) 180 mg/day, and rifampicin (RFP) 600 mg/day, with a target VCM trough level of 15-20 μg/mL, which was maintained within this range. Repeated TEE assessments during the treatment course revealed vegetation regression, no progression of valve destruction, and no overall clinical stability. Consequently, antibiotic therapy was continued without surgical intervention.

On hospitalization day 11, all three sets of blood cultures tested negative. However, the patient developed a lower leg rash and fever after VCM administration, thus suggesting a drug-induced reaction that prompted VCM discontinuation. Alternatively, LZD 1,200 mg/day was initiated on day 12 (Fig. 4). The following day, the platelet count decreased to 53,000 /μL. RFP-induced thrombocytopenia was suspected in the early phase of LZD treatment, leading to RFP discontinuation, and the LZD dose was reduced to 900 mg/day. With the patient's consent, TDM of the LZD plasma concentrations was conducted. Rapid HPLC (LM1010, Hitachi High-Tech Science, Tokyo, Japan) was used for measurements, targeting a trough concentration of 2-8 mg/L. The LM1010 system can rapidly measure drug concentrations after blood sampling and help clinicians optimize LZD concentrations in a timely manner. GM was administered as planned 14 days before discontinuation. On Day 25, the patient's platelet count remained within the normal range (211,000 /μL), while the LZD trough concentration was 7.98 mg/L - within the therapeutic window, but relatively high. By Day 31, the platelet count had decreased to 67,000 /μL, prompting a dose reduction of LZD to 600 mg and temporary discontinuation on days 32 and 33 based on the previously elevated trough level. On Day 34, the platelet count further declined to its nadir of 55,000 /μL; however, the LZD trough level measured that day had markedly decreased to 0.28 mg/L. In light of this substantial reduction in drug exposure, platelet recovery was anticipated and LZD was promptly resumed at 600 mg/day. Thereafter, the trough levels remained within the range of 2.18-5.51 mg/L, and the platelet count stabilized from 70,000-153,000 /μL, with no significant decline. After confirming negative blood cultures, intravenous antibiotics were administered for six weeks. On day 14 after antibiotic initiation, new cortical infarcts were observed; however, the patient remained asymptomatic. However, after completion of the six-week antibiotic regimen, no new infarctions were observed (Fig. 5). TEE confirmed the almost complete regression of vegetation and no worsening of MR (Fig. 6). The patient was discharged on day 53 and remained asymptomatic without evidence of recurrence for ＞ one year.

Figure 4.

Clinical course, antibiotic regimen, linezolid trough concentrations, and platelet count trends. The x-axis represents the number of days since the initiation of antibiotic therapy; platelet counts (black line) and linezolid trough concentrations (white bars). These findings suggest a temporal relationship between changes in the linezolid trough levels and subsequent platelet counts, with elevated trough levels preceding a decrease in the platelet count, and reduced trough levels preceding platelet recovery. After the linezolid dose is reduced to 600 mg, trough levels stabilized within the therapeutic range of 2-8 mg/L, thus maintaining sufficient efficacy for the treatment of infective endocarditis. GM: gentamicin, LZD: linezolid, RFP: rifampicin, VCM: vancomycin

Figure 5.

Head diffusion-weighted magnetic resonance imaging. (a) On day 14 after initiation of antibiotics, new cortical infarcts (arrows) are observed. (b) After completion of a six-week antibiotic regimen, no additional new infarctions are evident.

Figure 6.

Transesophageal echocardiography of the mitral valve performed after completing a six-week antibiotic treatment. (a) Two-dimensional 52° view and (b) three-dimensional en-face view of the mitral valve. The images demonstrate the almost complete resolution of vegetations on the prosthetic mitral valve ring. Mitral regurgitation has not worsened. LA: left atrium, LV: left ventricle

Discussion

This case involved subacute IE with vegetation on the prosthetic ring of the mitral valve caused by MRSE that developed several years after mitral valvuloplasty. LZD was administered as an alternative treatment for VCM intolerance. TDM-based dose adjustments effectively managed thrombocytopenia, allowing the patient to complete the treatment successfully.

The diagnosis took several months because of the atypical presentation. Although fever is reported in 86-96% of IE cases (1,2), the patient remains afebrile and exhibits no other suggestive signs, excluding recurrent cerebral infarctions, making the diagnostic process challenging. IE after mitral valvuloplasty is relatively rare, with varying prevalence rates during the chronic phase. Notably, approximately 50% of the cases caused by coagulase-negative staphylococci occur within the first year after valve surgery. Identified risk factors include incomplete valve repair and immunosuppression (2,4). The patient had a moderate MR due to recurrent mitral valve prolapse. The regurgitant jet may have impacted the prosthetic mitral ring, thus inhibiting endothelization and increasing susceptibility to vegetation formation. Although the patient tested positive for PR3-ANCA, no physical examination findings suggested vasculitis. As approximately 24% of IE cases are ANCA-positive, particularly in subacute IE (5), we considered ANCA positivity to be part of the IE disease process. Another challenge in this case was delayed bacterial growth, necessitating genetic testing to identify causative organisms. IE caused by coagulase-negative staphylococci rarely follows an acute course and the symptoms are often nonspecific (6). The absence of typical symptoms in this patient may have been due to biofilm formation on the prosthetic ring. Although MRSE does not typically exhibit slow growth, biofilm production may reduce its growth rate, further complicating early detection.

LZD is not formally approved for the treatment of IE in Japan and it is not recommended as a first-line agent in current guidelines (1,2). Although daptomycin is typically considered the first-line alternative to VCM, LZD has shown high efficacy in selected cases and may be considered in cases of VCM intolerance or treatment failure (7,8). LZD offers several advantages over daptomycin, including high oral bioavailability and excellent tissue penetration, with potential efficacy even in cases involving central nervous system complications (7-9). These properties make LZD a viable option for prolonged therapy when oral administration is necessary, particularly in the context of slow-growing infections that require extended treatment durations. In the present case, recurrent cerebral emboli were attributed to vegetation on a mitral annuloplasty ring, and the initial selection of LZD was based on its potential for effective long-term management. However, the infection ultimately responded more promptly than expected, thus suggesting that extended therapy might not have been necessary.

In this case, although thrombocytopenia developed, TDM-based adjustment of the LZD dosage enabled safe continuation of therapy without bleeding complications or the need for transfusion while maintaining appropriate drug levels for long-term use. Thrombocytopenia is a relatively common adverse effect of LZD, particularly in patients with prolonged treatment (＞14 days), advanced age, renal dysfunction, or elevated trough concentration (3). Cojutti et al. demonstrated that thrombocytopenia was independently associated with both the baseline platelet count and median LZD trough levels in a multivariate linear regression analysis (10). Elderly patients are particularly susceptible to IE and may require LZD therapy (2,11). However, the standard dose (1,200 mg/day) may be high in most elderly patients, thereby increasing the risk of thrombocytopenia (12). Previous studies have suggested that adjusting the LZD dosage based on blood concentration monitoring and maintaining trough levels between 2 and 8 mg/L could help prevent thrombocytopenia and allow treatment to continue beyond 28 days (10,13). In the present case, the platelet count dropped to 67,000 /μL six days after the LZD trough concentration reached 7.98 mg/L, which approached the upper limit of the therapeutic range and is consistent with previous reports linking elevated trough levels to hematologic toxicity. Although a trough level was not obtained precisely on day 32, the prior upward trend in drug concentrations informed our early clinical decision to reduce and temporarily discontinue LZD on Days 32 and 33. Subsequent confirmation of a markedly reduced trough level to 0.28 mg/L (day 34) allowed us to resume therapy without delay, thus maintaining antimicrobial efficacy, while minimizing the risk of severe thrombocytopenia. The serum LZD concentrations were measured using the LM1010 system, which is a rapid HPLC analyzer. This system simplifies HPLC procedures and provides stable and reproducible results with a minimal operator burden (14). Notably, LM1010 demonstrated a good correlation with conventional liquid chromatography-mass spectrometry methods and minimal inter-instrument variability. Compared to conventional methods, which typically require longer processing times, LM1010 can deliver results in approximately 7 min per sample, thus significantly reducing the turnaround time (15). This enables near-real-time TDM, facilitating prompt dose adjustments to balance antimicrobial efficacy with the risk of hematologic toxicity. Although this system may offer substantial clinical benefits in prolonged LZD therapy for IE, its widespread implementation in Japan is currently limited due to the lack of national healthcare reimbursement. Further studies are required to validate the efficacy of rapid HPLC-based TDM for LZD therapy and establish its role in clinical practice.

Conclusions

We successfully treated a patient with afebrile IE caused by slow-growing MRSE by using rapid HPLC-based TDM to guide LZD dose adjustments. This approach allowed for the completion of antimicrobial therapy while preventing thrombocytopenia and maintaining treatment efficacy. Our findings highlight the potential use of rapid HPLC-based TDM to optimize LZD therapy for IE.

Written informed consent was obtained from the patient for the submission and publication of this case report.

Author’s disclosure of potential Conflicts of Interest (COI).

Hiromichi Suzuki has received research funding from Hitachi High-Tech Science. The remaining authors declare no conflicts of interest.