Susceptibility to Cryptococcus neoformans Infection with Bruton’s Tyrosine Kinase Inhibition

ABSTRACT

Patients receiving the Bruton’s tyrosine kinase (BTK) inhibitor ibrutinib have an increased likelihood of fungal infections. The objectives of this study were to determine if Cryptococcus neoformans infection severity was isolate dependent with BTK inhibition and whether blocking BTK impacted infection severity in a mouse model. We compared four clinical isolates from patients on ibrutinib to virulent (H99) and avirulent (A1-35-8) reference strains. BTK knockout (KO) and wild-type (WT) C57 mice and WT CD1 mice were infected by intranasal (i.n.), oropharyngeal aspiration (OPA), and intravenous (i.v.) routes. Infection severity was assessed by survival and fungal burden (CFU per gram of tissue). Ibrutinib (25 mg/kg) or vehicle was administered daily through intraperitoneal injections. In the BTK KO model, no isolate-dependent effect on fungal burden was observed, and infection severity was not significantly different from that of the WT with i.n., OPA, and i.v. routes. Ibrutinib treatment did not impact infection severity. However, when the four clinical isolates were compared to H99, two of these isolates were less virulent, with significantly longer survival and reduced rates of brain infection. In conclusion, C. neoformans infection severity in the BTK KO model does not appear to be isolate dependent. BTK KO and ibrutinib treatment did not result in significantly different infection severities. However, based on repeated clinical observations of increased susceptibility to fungal infections with BTK inhibitor therapy, further work is needed to optimize a mouse model with BTK inhibition to better understand the role that this pathway plays in susceptibility to C. neoformans infection.

INTRODUCTION

Cryptococcus neoformans infection can cause devasting cases of meningoencephalitis, leading to long-term morbidity and mortality. Over the past 30 years, a subtle shift in the epidemiology of this infection has taken place, from HIV-infected patients to other immunocompromised populations, including patients with cancer or solid organ transplants (1). Among patients with cancer, patients with hematologic malignancies are more likely to develop C. neoformans infection than patients with solid tumor malignancies and are more likely to present with disseminated infection (2).

Ibrutinib, a small molecular inhibitor used in the treatment of B-cell malignancies, is a common irreversible inhibitor of Bruton’s tyrosine kinase (BTK) that decreases malignant B-cell proliferation and survival (3). BTK inhibitors have revolutionized the treatment of B-cell malignancies, improving patient outcomes particularly for elderly patients over the past decade (3, 4). Opportunistic fungal infections have been associated with the use of BTK inhibitors. However, prior clinical observations of patients with X-linked agammaglobulinemia, which results in a genetic deficiency of BTK, demonstrate that opportunistic fungal infections are rare (5–8).

In 2016, we observed two unusual cases of C. neoformans infection in patients with hematologic malignancies receiving ibrutinib (9). In these cases, both patients presented with disseminated C. neoformans infection within 4 weeks of starting ibrutinib, had minimal inflammation of the cerebrospinal fluid (CSF), and still suffered devastating morbidity from this infection with prolonged hospital courses. In fact, one of the patients died due to this infection. Since this early report, additional cases of patients developing invasive fungal infections while on ibrutinib have been reported for other fungal pathogens (5, 10, 11).

We sought to characterize the impact of BTK inhibition on susceptibility to C. neoformans infection through the following objectives: (i) impact of clinical isolates on infection severity in BTK gene knockout (KO) (Btk^−/−) mice compared to wild-type (WT) (Btk^+/+) mice, (ii) impact of BTK KO on infection severity, (iii) impact of BTK KO on survival with C. neoformans infection, and (iv) impact of BTK pharmacologic inhibition by ibrutinib on infection severity.

RESULTS

Clinical cases.

The four clinical cases of C. neoformans infection in patients on ibrutinib are depicted in Table 1. These cases are notable for early onset of infection following initiation of ibrutinib (median, 37 days), presentation with disseminated infection, and poor survival postinfection (median, 313 days). In the two cases in which there was brain involvement (JPC6193 and JPC6195), the patients had minimal inflammation within the CSF (only three nucleated cells in each patient’s sample) and negative CSF Cryptococcus antigen titers (IMMY, Norman, OK) despite growth of C. neoformans in CSF fungal culture.

TABLE 1.

Clinical characteristics of patients who developed C. neoformans infection while on ibrutinib therapy at Duke University Medical Center^a

Patient isolate identifier	Underlying disease	Therapies prior to ibrutinib	Time from ibrutinib start to infection diagnosis (days)	Absolute lymphocyte count at time of infection	Infection sites	CSF study findings	Clinical outcome
JPC6192	Marginal zone lymphoma	BR, R-CHOP	18	1.06/L	Bloodstream, lung	NA	Deceased due to complications of relapsed lymphoma, 714 days postinfection diagnosis
JPC6193	Indolent lymphoplasmacytic lymphoma	Rituximab, BR	45	0.7/L	CSF, lung	Opening pressure, 7 cm H2O; fungal culture, C. neoformans (1 colony); CSF CrAg negative; 3 nucleated cells/μL; protein, 43 mg/dL; glucose, 57 mg/dL	Deceased due to complications of relapsed lymphoma, 553 days postinfection diagnosis
JPC6194	Mantle cell lymphoma	R-CHOP, BR, R-GEM, R-DHAP, R-HyperCVAD	137	0.6/L	Bloodstream, skin, lung	NA	Deceased due to complications of relapsed lymphoma, 72 days postinfection diagnosis
JPC6195	Chronic lymphocytic leukemia	FCR	29	59.2/L	Bloodstream, lung, CSF	Opening pressure not recorded; fungal culture, C. neoformans; CSF CrAg negative; 3 nucleated cells/μL; protein, 30 mg/dL; glucose, 157 mg/dL	Deceased due to complications from septic shock from C. neoformans infection, 34 days postinfection diagnosis

^aBR, bendamustine and rituximab; R-CHOP, rituximab, cyclophosphamide, hydroxydoxorubicin, vincristine (trade name Oncovin), and prednisone; R-GEM, rituximab and gemcitabine; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin; R-HyperCVAD, rituximab, cyclophosphamide, vincristine, adriamycin, and dexamethasone; FCR, fludarabine, cyclophosphamide, and rituximab; CSF, cerebrospinal fluid; CrAg, cryptococcal antigen; NA, not applicable.

Animal experiment 1: are BTK KO mice more susceptible to C. neoformans, and does infection severity vary based on clinical Cryptococcus isolate?

Our first experiment attempted to address whether the BTK KO mice would have more severe fungal disease, evidenced by higher fungal burden, and whether clinical isolates from patients on ibrutinib with disseminated cryptococcal disease produced similar levels of disease in WT and BTK KO mice. Four clinical isolates were compared to H99 and A1-35-8 controls in WT and BTK KO mice. Strain A1-35-8 had been shown to produce minimal disease in mice, whereas H99 caused severe disease and death when given to mice. The four clinical isolates had not been previously tested in mice or genotyped prior to the completion of this study. We utilized the intranasal (i.n.) route of infection based on a previous study of the impact of the BTK target on invasive fungal infection (12). Mice were sacrificed on day 7 postinfection to evaluate the fungal burden in the lungs and brain (N = 24 WT, 67 BTK KO). We did not observe a significant difference in lung or brain fungal burden based on C. neoformans isolate or WT versus BTK KO genotype (Fig. 1).

FIG 1.

Impact of clinical Cryptococcus neoformans isolates on infection severity in BTK KO model versus WT. Four clinical isolates from patients who developed C. neoformans infection while on ibrutinib are compared to H99 and A1-35-8 controls in C57 BTK KO versus WT mice (i.n. route of infection). Infection severity was measured by lung (A) and brain (B) fungal burden. We found no significant difference in lung or brain burden based on C. neoformans isolates (N = 67 BTK KO versus 24 WT; P > 0.05).

For the next portion of this experiment, we utilized oropharyngeal aspiration (OPA) infection as we wanted to ensure that the yeasts reached the lung tissue and used H99 as the only infection strain. To compare infection severities between BTK KO and WT mice, we performed OPA infection (N = 15 WT, 20 BTK KO) and sacrificed mice on day 14 (Fig. 2A and B). We harvested lung tissue from nine mice from this experiment (N = 4 WT mice, 5 BTK KO) and did not observe a significant difference in histopathology by histologic scoring (Fig. 3). We then performed a survival experiment through a 28-day endpoint (N = 12 WT, 17 BTK KO) (Fig. 2C), which did not demonstrate a significant difference in survival based on genotype.

FIG 2.

Infection severity in BTK KO versus WT mice. We utilized oropharyngeal infection (A to C) and tail vein infection (D) with H99. We found no significant difference in fungal burden or 28-day survival with H99 infection in BTK KO versus WT mice. (A) Lung fungal burden 10 days postinfection with H99 via OPA route of infection (N = 15 WT, 20 KO; P = 0.57); (B) brain fungal burden 10 days postinfection with H99 via OPA route of infection (N = 15 WT, 20 KO; P = 0.058); (C) 28-day survival postinfection with H99 via OPA route of infection (N = 12 WT, 17 KO; P = 0.89); (D) brain fungal burden 48 h postinfection with H99 via tail vein route of infection (N = 10 WT, 8 KO; P = 0.093).

FIG 3.

(A) Histology scores for day 14 postinfection for lung tissue necrosis (left), hemorrhage (middle), and edema (right) from BTK KO and WT mice. All mice were infected with H99 by OPA and sacrificed on day 10 postinfection. There was no significant difference in histology scores of necrosis, hemorrhage, or edema between BTK KO and WT mice (N = 4 BTK, 5 WT; P > 0.05). (B) Representative photographs of histology (hematoxylin and eosin staining) from BTK KO model (i, ii) and WT (iii, iv) mice. The numbers 1 to 4 designate representative examples: 1, encapsulated Cryptococcus; 2, hemorrhage; 3, inflammatory cell infiltration and necrosis; and 4, edema.

Reflecting upon the two clinical cases of C. neoformans infection with rapid brain involvement in these patients taking ibrutinib, we hypothesized that susceptibility to CNS infection with BTK inhibition might be due to an impaired blood-brain barrier. To test this hypothesis, we performed tail vein injections with strain H99 (N = 10 WT, 8 BTK KO), bypassing the lungs, to analyze brain yeast burden (Fig. 2D). Mice were sacrificed at 48 h postinfection. We did not observe a significant difference in brain yeast burden between WT and BTK KO mice.

Next, we decided to reevaluate the impact of isolates from the ibrutinib clinical cases on infection severity in outbred CD1 mice (N = 25 mice, 5 mice per isolate). Mice were observed through day 57 post-OPA infection. Three of the four clinical isolates, JPC6192, JPC6194, and JPC6195, were associated with significantly longer survival than that of strain H99 (Fig. 4) (P values = 0.0220, 0.0017, and 0.0017, respectively). Interestingly, one clinical isolate, JPC6195, did not cause any mouse deaths and failed to infect the brain. In contrast, this isolate was associated with the most severe clinical presentation of infection in its human host, leading to septic shock, intubation, and death within 1-month postinfection diagnosis. JPC6192 and JPC6193 caused brain infection in all mice, while JPC6194 caused brain infection in two of five mice.

FIG 4.

Differential mortality induced by Cryptococcus neoformans clinical isolates in CD1 mice. We analyzed survival and brain fungal burden in CD1 mice infected with the four clinical isolates from patients infected with C. neoformans while on ibrutinib, using H99 as a reference strain (N = 25 mice; 5 mice per infection isolate). Three of the four clinical isolates, JPC6192, JPC6194, and JPC6195, were associated with significantly longer survival than that for H99 (P < 0.0001).

Animal experiment 2: does treatment of outbred wild-type CD1 mice with ibrutinib increase infection severity?

As we did not find a significant difference in fungal burden or survival between WT and BTK KO mice with acute infection, we next hypothesized that human susceptibility to C. neoformans infection may be due to an off-target effect of ibrutinib rather than direct BTK inhibition. CD1 mice were infected with either H99 or clinical isolate JPC6195 (N = 40; N = 10 mice treated with ibrutinib and infected with H99, 10 mice treated with vehicle and infected with H99, 10 mice treated with ibrutinib and infected with JPC6195, and 10 mice treated with vehicle and infected with JPC6195). Mice were then sacrificed on postinfection day 10 to determine yeast burden. We did not observe a significant difference in lung or brain fungal burden based on ibrutinib versus vehicle treatment (Fig. 5). However, we did confirm that OPA infection with H99 resulted in significantly higher lung and brain fungal burden than did infection with JPC6195 in CD1 mice.

FIG 5.

Impact of ibrutinib treatment on infection severity with H99 and JPC6195 infection in CD1 mice. We compared lung (A) and brain (B) fungal burdens with H99 and JPC6195 infection in CD1 mice treated with ibrutinib versus vehicle (N = 10 JPC6195 + ibrutinib, 10 JPC6195 + vehicle, 10 H99 + ibrutinib, and 10 H99 + vehicle). We found no significant difference in fungal burden in mice with ibrutinib versus vehicle treatment (P > 0.05). We confirmed that that H99 infection resulted in significantly higher lung and brain fungal burden than did infection with JP6195.

DISCUSSION

In this study, we have taken a clinical observation of increased susceptibility to C. neoformans infection in humans on ibrutinib and applied it to a mouse model of acute infection but were unable to recapitulate this finding. The clinical cases of C. neoformans infection in patients on ibrutinib are notable for early onset of infection following initiation of ibrutinib and poor survival postinfection. Multiple earlier studies have demonstrated that patients with lymphoid malignancies are at risk for developing C. neoformans infection, particularly with cytotoxic chemotherapy such as fludarabine and cyclophosphamide (13–16). However, the majority of these studies describing the epidemiology of C. neoformans infection took place prior to the U.S. Food and Drug Administration approval of ibrutinib in the 2010s. Our cases of C. neoformans infection following ibrutinib therapy illustrate a critical need to understand the pleiotropic immune deficits introduced through BTK inhibition and their impact on susceptibility to C. neoformans infection.

However, in the present study, the BTK KO model did not have significantly different infection severity from that of the WT with i.n., OPA, and early CNS entry tail vein routes based on tissue fungal burden or mouse survival. In addition, ibrutinib administration at a dose replicating human exposure did not appear to impact infection severity in CD1 mice.

In contrast to our findings, other investigators have noted increased susceptibility to fungal infections with Btk mutations in mouse models. One study utilized X-linked immunodeficient (XID) mice, which have a mutation in Btk, and two infection strains, H99 as a virulent control strain and CN52D as a chronic infection strain (12). The investigators noted higher lung and brain fungal burdens than those of control mice at 3 and 6 weeks postinfection for H99 and CN52D strains, respectively. Additionally, XID mice infected with CN52D had impaired alveolar macrophage phagocytosis and a disorganized lung inflammatory response by histology. Based on these findings, the investigators concluded that therapies targeting the BTK pathway may lead to increased human susceptibility to C. neoformans infection. Of note, the XID mice have a spontaneous X-linked mutation in Btk, whereas the BTK KO model possesses a gene-targeted mutation in embryonic stem cells leading to the inability to detect Btk gene products (17).

A study in a BTK KO mouse model of pulmonary aspergillosis revealed significantly reduced 14-day survival, greater weight loss, more severe lung damage, and higher lung fungal burden in BTK KO mice than in WT mice (18). These results support the findings from the XID mouse model that the BTK pathway can play a role in host response to fungal infections.

Regarding the impact of clinical isolates on C. neoformans infection severity, we had hypothesized that infection isolates might have been uniquely adapted to be successful in the chronic lymphoid malignancy patient population. However, based on our initial experiment with i.n. infection using four clinical isolates and two control strains, C. neoformans infection severity in the BTK KO model did not appear to be isolate dependent. However, after performing a follow-up survival experiment comparing the impacts of isolates on infection severity in CD1 mice, we had differing results. In fact, utilization of multiple clinical isolates did show variation in infection severity based on two isolates being associated with significantly improved survival and less frequent brain infection. Interestingly, JPC6195, the clinical isolate associated with the worst human outcome, showed an impressive lack of correlation between mouse and human outcome, causing no brain infection or death in CD1 mice.

Our results clearly emphasize the complexity of studying targeted therapies such as small molecule inhibitors and monoclonal antibodies on risk for infection. These drugs are potent inhibitors of cellular processes to treat the underlying disease and its immune dysregulation. Despite the success in using mouse models to dissect virulence traits in C. neoformans infection, the mouse is not a clear validating model for human disease in this case. In order to understand mechanisms with animal models, it may be necessary to adjust the complexity of the environment to replicate the impact of the host on the fungal pathogen. Ibrutinib’s undesirable effects of weakening antifungal defenses are most likely conditional and require a very particular set of biological conditions, which the present study has not reproduced. Notably, the present study’s findings support the general safety of ibrutinib with the exception of rare and undefined conditions that need to coexist for the undesirable effects of weakened antifungal response to become significant.

Moving forward, there are at least two things to consider. First, the murine model that we used is an acute infection, and BTK blockage might impact a more chronic or subacute disease process with impact of reactivation. Therefore, we need to use cryptococcal isolates that, in mice, produce chronic infections. Second, our present murine models do not replicate the underlying hematologic malignancies, which are so critical in the host response. Our future goal is to utilize a mouse model of chronic lymphocytic leukemia, a disease for which BTK inhibitors are used for first-line therapy, to better analyze the effect of BTK inhibitors on C. neoformans infection.

MATERIALS AND METHODS

Clinical data acquisition.

Patients who developed C. neoformans infection while taking ibrutinib were identified through inpatient consultation on the Duke Transplant Infectious Diseases service. Clinical data were retrieved upon review of the Duke electronic medical record. This study received approval and a waiver of informed consent from the Duke Institutional Review Board.

Yeast isolates and growth conditions.

Four clinical isolates from patients who developed C. neoformans infection while on ibrutinib were retrieved from the Duke Clinical Microbiology Laboratory (labeled JPC6192, JPC6193, JPC6194, and JPC6195). Clinically virulent reference (H99) and avirulent reference (A1-35-8) strains were also used. Cryptococcus isolates were cultured from original freezer stock onto yeast extract-peptone-dextrose (YPD) growth agar and grown for 3 to 4 days at 30°C. A single colony was inoculated into 5 mL of YPD agar and grown overnight at 30°C on a rotary shaker incubator. Cells were harvested and washed twice with 10 mL of phosphate-buffered saline (PBS) and resuspended in PBS (concentration, 10⁶ CFU/mL).

Animal studies.

All animal experiments were conducted in accordance with a protocol approved by the Duke University Institutional Care and Use Committee. BTK KO C57 breeding pairs were received from M. S. Lionakis from the National Institutes of Health and expanded at the Duke Breeding Core. Genotypes were confirmed by PCR by a third-party vendor (Transnetyx, Inc., Cordova, TN). A combination of female knockouts and male hemizygous knockouts aged 7 to 14 weeks were used for the experiments. Male CD1 mice, 8 to 10 weeks old and weighing 22 to 24 g, were purchased from Charles River Laboratories (Wilmington, MA). For i.n. infections, mice were briefly sedated in 1 to 3% isoflurane, and then 20 μL of inoculum containing 5 × 10⁴ cells was instilled into the nares. The mice recovered and were returned to their home cage. For OPAs, the mice were anesthetized with 2 to 5% isoflurane and then suspended vertically by their incisors. The tongue was extended with forceps, and 50 μL of inoculum was instilled into the oropharynx using a pipette. The mice were allowed to aspirate the fluid and then returned to their home cage. For tail vein injections, the mice were secured in a restrainer (Kent Scientific) and then placed under a supplemental heat source for 1 min. The tail was then extended, and 0.1 mL of the inoculum containing 5 × 10⁴ cells was injected. The mice were then returned to their home cage. Mice were monitored daily and euthanized if they had ≥15% body weight loss or if symptoms of infection were observed.

Ibrutinib was purchased from MedChemExpress (Monmouth Junction, NJ), and suspension was made in dimethyl sulfoxide (DMSO; 100 mg/mL) and diluted in 0.4% methyl cellulose. Vehicle control contained DMSO and 0.4% methyl cellulose. We treated the mice with daily intraperitoneal injections of ibrutinib (or vehicle) starting 7 days prior to infection and through the duration of the experiment.

Fungal burden.

Lungs and brains of mice were harvested and homogenized by bead milling the tissue for 30 s in 1 mL of PBS. Homogenized tissue was then serially diluted and plated onto YPD agar containing 100 μg/mL chloramphenicol. Plates were incubated for 2 to 3 days at 30°C, and colonies were counted as CFU/g of tissue weight.

Histopathology.

We assessed histopathology using hematoxylin and eosin staining and histologic scoring of necrosis, hemorrhage, edema, and inflammation, per the Duke Veterinary Diagnostic Laboratory (Division of Laboratory Animal Resources), as previously described (19). Mice were euthanized, and the thorax was opened to expose the lungs. A metal cannula was inserted into the trachea and secured with surgical thread. Formalin was instilled into the lungs at a height of 25 cm for 10 min. The lungs were removed, submerged in formalin for 24 h, and then switched to 70% ethanol. They were paraffin embedded, sectioned at 5 μm, and then stained with hematoxylin and eosin. The histology was then scored by a veterinary pathologist for necrosis, hemorrhage, and edema.

Statistical analysis.

P values were computed using the Kruskal-Wallis test for group effect and the Wilcoxon rank sum test for within-group comparisons, or using a Student’s t test. P values were adjusted using the Bonferroni-Holm method when multiple comparisons were performed. A P value less than 0.05 was considered statistically significant. Survival at 28 days was analyzed by Kaplan-Meier curves and compared by log rank test. Histology scores were compared by the Wilcoxon signed rank test. R studio (version 4.2.1) was used for all analyses.