Bloodstream infection by Lactobacillus rhamnosus in a haematology patient: why metagenomics can make the difference

Bloodstream infections (BSI) are a constant threat to hospitalized patients due to various risk factors associated with underlying diseases such as cardiovascular disease, cirrhosis, diabetes, and cancer [1], as well as treatment-related factors like exposure to immunosuppressive therapies and the use of indwelling devices [2]. These infections, along with sepsis, represent a leading cause of mortality in patients with hematologic malignancies or solid tumors undergoing intensive cytotoxic chemotherapy, particularly in those with acute leukemia who experience chemotherapy-induced neutropenia [3, 4].

In addition to the risks related to neutropenia post-cytotoxic therapy, hospitalization and inpatient management come with other important risks to the hematologic patient. It is widely known that extensive use of antibiotics confers risk of antibiotic-related adverse events, including antibiotic resistance and toxicity [1] and that long-term hospitalization favors colonization and infection by multidrug resistant strains [2]. What is less well described is the role of the gut microbiota in the pathogenesis of treatment-related morbidity and mortality in haematology inpatients.

The gut microbiota is engaged in a complex relationship with the mucosal epithelium, intestinal epithelial cells, and underlying immune tissues, establishing a dynamic equilibrium that confers resistance to infection and allows the absorption of nutrients [3, 4].

Exogenous factors in the inpatient setting, such as antibiotic use, chemotherapy, and immunosuppressive therapy, can disrupt intestinal microbial equilibrium, resulting in the loss of commensal diversity and in the consequent expansion of potentially pathogenic bacteria (pathobionts [5, 6]. This variation reflects the breakdown of ecological mechanisms that normally constrain pathobionts, including niche competition, colonization resistance, and mutualistic interactions. The depletion of beneficial anaerobes reduces competition for nutrients and space, while also impairing the production of antimicrobial metabolites and immune-modulating signals that maintain mucosal integrity [7]. Finally, immune system impairment further exacerbates this imbalance by altering the metabolic environment, for instance by increasing the availability of oxygen or nitrate, which selectively allows the outgrowth of facultative anaerobes such as Enterobacteriaceae [8, 9].

Radiation and chemotherapy, which constitute the conditioning regimen for hematopoietic stem cell transplantation can promote a damage of the gut epithelial stem cells and weaken innate immune defences by affecting host-microbe interactions [3]. Moreover, long-term antibiotic treatment causes a significant change in gut microbiota favouring the overgrowth of pathogens or pathobionts in a process known as intestinal domination [3, 10, 11], that is commonly recognized when at least 30% of the microbiota by a single predominating bacterial taxon occurs [3]. Recent studies have shown that the increased relative abundance of specific taxa in the gut bacterial communities enhances the risk of bacteremia in certain settings [3, 12, 13]. Analysis of gut microbiota composition and bacterial genotype may represent a promising tool to assess infection risk in colonized patients [14], although further studies are needed for their use in clinical application [15, 16]. While the effects of intestinal domination of known pathobionts, such as K. pneumoniae, have been decoumented [14]; the potential impact of infections caused by commensals has not been described. Lacticaseibacillus spp., formerly known as Lactobacillus spp., are a Gram-positive component of human mucosal microflora, including the oral, vaginal and intestinal flora [17‐20]. As a result of their beneficial properties, isolates of these species are commonly used as probiotics [17, 21, 22]. When their equilibrium with other microbial components is disrupted, Lacticaseibacilli may morph into opportunistic bacteria, giving rise to clinically invasive infections in immunocompromised patients, including abscesses, infectious endocarditis and sepsis [23, 24].

Over the years, a number of BSIs by Lactobacillus rhamnosus (L. rhamnosus) have been described (Table 1). While the use of probiotics is generally cautious and limited in immunocompromised or critically ill patients, especially in Western healthcare systems, they are occasionally administered as adjuncts, for instance to prevent antibiotic-associated diarrhoea [25, 26]. L. rhamnosus has been implicated as an emerging pathogen in patients receiving probiotics, including cardiosurgical patients [27], diabetic patients [28] and in infants [29], raising concerns about their safety [30]. Despite clinical suspicion, no definitive mechanistic explanation for gut translocation or the origin of bloodstream infection has yet been established [31]. In this study, we provide experimental evidence of intestinal domination by L. rhamnosus, which was subsequently identified in the bloodstream.

A 20-year-old female was referred to hematology center for CAR-T cell therapy following relapse of Philadelphia-negative B-cell acute lymphoblastic leukemia (ALL). The patient was initially diagnosed with ALL in March 2022 at another institution after presenting with lower limb bruising and marked leukocytosis (WBC 63,710/mm³) with 80% peripheral blood blasts. She received standard multi-agent chemotherapy (LAL1913 protocol), which included an initial steroid and cyclophosphamide prephase, followed by induction therapy with vincristine, idarubicin, peg-asparaginase, and dexamethasone, along with prophylactic intrathecal methotrexate, cytarabine, and prednisone until April 2022. The patient subsequently underwent seven additional cycles of induction and consolidation therapy until December 2022, achieving measurable residual disease (MRD) negativity. Maintenance therapy began in January 2023 and continued until January 2024, when she experienced her first relapse.

Upon relapse confirmation (85% blasts on bone marrow morphology; CD22 positivity confirmed by flow cytometry), the patient received a single cycle of Inotuzumab as a bridge to CAR-T, once again achieving MRD negativity by flow cytometry. Following leukapheresis for CAR-T manufacturing, a second Inotuzumab cycle was administered, with MRD negativity confirmed via flow cytometry and molecular testing. The patient received CAR-T cell infusion in April 2024, following standard lymphodepleting chemotherapy with fludarabine and cyclophosphamide. Her post-CAR-T course was uneventful, and she remained in molecular remission until July 2024, when MRD re-emerged (10⁻³).

In response to molecular relapse, she was started on blinatumomab therapy, which was initially complicated by neurotoxicity. Treatment was temporarily halted, and the patient received antiepileptic therapy and high-dose corticosteroids. Blinatumomab was then resumed on an outpatient basis as a bridge to transplant, along with intrathecal chemotherapy, steroids, and weekly vincristine. This outpatient period was further complicated by SARS-CoV-2 infection, treated with nirmatrelvir.

The patient was admitted for allogeneic hematopoietic stem cell transplantation in early October 2024. A bone marrow aspirate at admission confirmed morphologic persistence of disease, prompting the administration of vincristine and corticosteroids. Due to persistent SARS-CoV-2 positivity on day 9 of admission, she received a course of remdesivir based on infectious disease consultation. Admission microbiological testing for gastrointestinal pathogens (bacteria, viruses, and parasites), fungi (germ tube test), multidrug-resistant organisms, and cytomegalovirus was negative.

Conditioning therapy commenced on day 26 and included total body irradiation (2 Gy twice daily on days − 8 to -6), fludarabine (40 mg/m² on days − 5 to -3), followed by infusion of peripheral blood stem cells from a 7/8 HLA-mismatched unrelated donor. Graft-versus-host disease (GvHD) prophylaxis included anti-thymocyte globulin (ATG), post-transplant cyclophosphamide (50 mg/kg on days + 3 and + 5), cyclosporine (from day + 5), and mycophenolate mofetil (1000 mg twice daily from day + 5 to + 30). A summary of the clinical course is presented in Fig. 1a.

Post-transplant complications included pulmonary aspergillosis (treated with isavuconazole and caspofungin) and HHV-6 encephalopathy (treated with foscarnet). Rhinovirus was also detected. Febrile neutropenia and catheter-related bloodstream infection (BSI) due to Staphylococcus epidermidis occurred during conditioning and were treated with meropenem, vancomycin, and liposomal amphotericin B. An Enterococcus faecalis infection was treated with cefepime, which was later discontinued due to neurotoxicity. Microbiological findings and inflammatory markers are presented in Fig. 1b.

In January 2025, the patient experienced additional severe transplant-related complications, including acute and chronic GvHD (maximum grade 3 cutaneous and gastrointestinal), for which she received corticosteroids, ruxolitinib, octreotide, and biweekly therapeutic plasma exchange. She also developed sinusoidal obstruction syndrome (treated with defibrotide) and transplant-associated thrombotic microangiopathy (TA-TMA), for which she received fresh frozen plasma infusions. Despite maximal therapeutic and supportive interventions, the patient died of multiorgan toxicity on day + 76 post-transplant. Beginning on day + 70 of admission, blood cultures drawn from peripheral veins, peripherally inserted central catheter (PICC), central venous catheter (CVC), and midline catheter were all positive for Lacticaseibacillus rhamnosus (L. rhamnosus). During this time, inflammatory markers associated with sepsis, including D-dimer and procalcitonin, were markedly elevated (Fig. 1b). On day + 70, aerobic and anaerobic blood culture bottles were incubated using the BACT/ALERT^® 3D system (bioMérieux, Marcy l’Étoile, France). Three aerobic cultures (one from PICC, two from CVC) became positive after 46 h, revealing Gram-positive rod-shaped bacteria. Identification using the MALDI BioTyper^® system (Bruker Daltonics, Bremen, Germany) confirmed L. rhamnosus with a score of 2.38 (reliable identification: >1.699). The second and third best matches also identified L. rhamnosus (scores: 2.31). Positive cultures were subcultured on Tryptic Soy Agar (TSA) and Columbia agar and incubated at 37 °C with 5% CO₂, confirming the identification.

The genome of the clinical isolate was sequenced (see Additional File 1: Methods). Upon detection of L. rhamnosus in blood cultures, a faecal sample was collected and subjected to 16 S rRNA gene-targeted gut microbiota profiling (see Additional File 1: Methods). As shown in Fig. 2a, sequencing of the V3-V4-V6 regions revealed highest relative abundances for L. paracasei (42.4%), Pediococcus acidilactici (18.7%), and unclassified Lactobacillaceae (16.8%). Given that 16 S sequencing cannot reliably differentiate species within the Lactobacillus casei group, shotgun metagenomic sequencing was performed. This analysis revealed L. rhamnosus as the dominant species (71.6%), followed by Pediococcus acidilactici (17.4%) (Fig. 2b). Taxonomically assigned reads corresponding to L. rhamnosus were extracted and assembled for genomic comparison with the clinical blood isolate. The two reconstructed genomes differed by only 18 single nucleotide polymorphisms (SNPs), strongly suggesting a clonal relationship. These findings demonstrate gut domination by L. rhamnosus and subsequent bloodstream infection, indicating a potential translocation event and highlighting the importance of intestinal microbiota surveillance in high-risk patients.