Clostridioides difficile recurrence in individuals with and without cancer: a Swedish population-based cohort study

Clostridioides difficile is a spore-forming bacterium that can be a member of the intestinal microbiome in healthy individuals [1, 2]. Yet, certain strains may release toxins, making them pathogenic [3, 4]. The human gut microbiome, comprising of bacteria, archaea, viruses, and eukaryotes, can prevent colonization and overgrowth, through direct and indirect mechanisms [1, 2, 5]. However, intestinal dysbiosis, also described as an “unhealthy” alteration of the gut microbiome, is a growing concern [1, 2, 4, 6]. C. difficile can contribute to this dysregulated gut microbiome, which may trigger the immune system and generate an environment favorable to overgrowth, colonization and infection [2]. In this case, the combination of ingestion of spores from the toxin-producing C. difficile, e.g., during hospitalization, and increased susceptibility (e.g., due to treatment with broad-spectrum antibiotics) can lead to C. difficile infection (CDI) [4]. CDI can lead to serious complications such as intense diarrhea, which can cause imbalances in electrolytes, dehydration, unstable blood flow, toxic megacolon, shock, and even mortality [7].

The incidence of CDI in adults and children remains high and is even increasing in several settings [8‐11]. As one of the most common healthcare-associated infections in the Western world, CDI is a major health burden and public health threat [7, 8, 12, 13]. According to an extensive meta-analysis including 57 studies on recurrence, approximately 10–20% experience recurrence particularly among hospital-acquired CDI, with a maximum recurrence of 64% [9]. Significant risk factors for CDI recurrence, based on three meta-analyzes, include antibiotic use, advanced age, proton pump inhibitor (PPI) use [14‐16], and renal insufficiency [14]. Several meta-analyzes confirmed the effect of antibiotics and PPIs in more detail confirming the increased CDI risk [17‐20] and recurrence risk [21‐23]. Hospitalization and gut dysbiosis are also reported as risk factors for recurrence but results are inconsistent [6, 9, 16].

Patients with cancer are considered particularly vulnerable to CDI, mainly because of their immunocompromised state, intense treatment, regular healthcare-contact, and higher overall susceptibility to infections [24‐26]. Several studies reported a higher incidence of CDI in patients with cancer than those without [27]. Whether this is also the case for recurrence seems less clear, as some studies do report cancer as a risk factor [28, 29], while others do not find significant associations, and effect estimated for cancer as risk factor which are clearly smaller than those related to exposure to antibiotics and PPIs [30‐33]. Exploring the risk of recurrence among different cancer types is challenging due to the heterogeneity of cancer types and treatments, yet it is of high clinical relevance, as some individuals may benefit from closer follow-up and adapted treatment if considered at high risk. If there are differences among cancer types, this could shed light, or at least open the discussion on potential pathophysiological mechanisms of CDI and recurrence, particularly to distinguish between re-infection and relapse [34, 35].

There is a lack of large-scale population studies that have investigated the risk of recurrence including community-acquired CDI, comparing all cancer types. Therefore, we conducted a large-scale Swedish population-based study, to investigate CDI recurrence in individuals with cancer (diagnosed within 1 year, and history of cancer) compared to individuals without cancer, and to examine which cancer types (by anatomical location) present with the highest recurrence risk.

The cohort included 43,150 individuals, with 3882 having ongoing cancer, 9725 with a cancer history and 29,543 having no cancer history (Fig. 1). Overall, 45.8% of the individuals were men, while 54.2% were women. Most individuals fell into the 65–84 age group (49.4%), followed by the ≥ 85 age group (25.4%) and the 0–64 age group (25.2%). The Charlson comorbidity score showed a varied health profile, with notable proportions having scores of 2 (18.7%) and 5 (26.7%). Most individuals acquired CDI in a hospital setting (91.6%), while 7.2% was community-acquired. Overall, most individuals had a recent history of antibiotic usage (97.1%) and a substantial portion had used proton pump inhibitors (71.5%) and non-steroidal anti-inflammatory drugs (59.0%) with limited differences between groups. The death rate before CDI recurrence within 8 weeks was 15.4%. The CDI recurrence within 8 weeks was 16.8% (Table 1).

This study based on one of the largest nationwide cohorts of individuals with CDI, is to our knowledge, the first population study comparing CDI recurrence between different cancer types with the highest odds of occurrence in patients with oral cancer and lowest in those with esophageal cancer. Our study was not restricted to active malignancies as our individuals with cancer could have been hospitalized for non-cancer related indications, and our main analyzes included all individuals with a cancer diagnosis during the last year. This may partially explain the apparent overall lower odds of recurrence in cancer patients, even after correcting for mortality, compared to those without cancer. Yet, one in five with a recent cancer diagnosis (< 1 year) or cancer history died within 2 months after CDI; compared to only 13% of individuals without cancer, which may be primarily due to the cancer or to the CDI or a combination of both. It does suggest CDI is a serious complication for those with cancer, an important contributing factor to death of individuals with cancer, and a not-negligible competing risk in our models.

Our apparent less frequent recurrence compared to individuals without cancer may seem contra-intuitive and contrary to some previous studies which have suggested increased recurrence [50‐52]. Yet, previous meta-analyzes did not find strong evidence for an association [14‐16], and other studies show similar results [12, 53]. The inclusion of community-acquired CDI in our study probably made little difference, as 97% of those with ongoing cancer was deemed health-care acquired compared to 90% of those without cancer.

The strengths of our study are the large population-based cohort of 43,150 individuals, which enhances the strength of the findings. Secondly, the inclusion of multiple cancer types, among individuals with an ongoing cancer or cancer history, allowed for a comprehensive analysis of the association between cancer type and CDI recurrence. An important limitation is that we could not incorporate (neo-)adjuvant cancer treatment and in-hospital drug use (including antibiotic prophylaxis when e.g., cytotoxic chemotherapy is administered), as these were not recorded in the registries. We also lack information on frailty, body mass index, weight loss, smoking or other residual confounders. Due to the registry nature, no information was available on specific C. difficile strains, which may pose different risks of recurrence e.g., due to different antimicrobial resistance patterns [6, 24], and may have helped to distinguish between reinfection and relapse [34, 35]. Underreporting and underdiagnosing of CDI, may also have contributed to a selection bias toward more severe cases. Our study encountered the challenge of limited existing literature, posing complexities during the design. Nevertheless, to ensure methodological integrity and avoid data-driven exploration, we followed a prior developed study protocol. This protocol defined CDI recurrence within an 8-week timeframe after the initial episode, considering deaths within this period as competing events. For the logistic regressions, our results excluding deaths may have created a selection bias, yet leaving them in underestimated the odds of recurrence [54]. Although the cause-specific hazard model was utilized, the presence of right-censored observations, which may not precisely reflect true right censoring from a statistical standpoint, introduces the possibility that the impact of deaths on CDI recurrence might not be adequately captured. The classification of cancer into ongoing cancer and cancer history presented challenges due to the wide variation in the clinical course of different cancer types. We approached this issue by considering that patients who survived the initial cancer episode would likely undergo a period of intense treatment, including surgery, chemotherapy, and/or radiotherapy, lasting less than a year. We hypothesized that individuals would be at the highest risk of CDI and recurrence during this active treatment phase. However, it is important to recognize that for many patients, cancer has long-term health effects. They may not achieve remission, experience complications from cancer or its treatment, or have a reduced overall quality of life, all of which could potentially impact their risk of CDI and recurrence and survival.

Although it is interesting to compare all cancer types, this comes with major challenges including difficulties to compare cancer aggressiveness, staging and treatment—and even our cut-off of 12 months since cancer diagnosis is not sensitive enough to distinguish between actively treated malignancies and those cured or in remission. Intuitively, we expected a higher overall recurrence, and we do expect more recurrences if we could more clearly define those with active cancer, as they will be more commonly hospitalized and exposed to antibiotics. Although almost all individuals with cancer in our cohort were categorized under healthcare associated CDI (97.1%), it could be considered to subcategorise cancer by prior healthcare consumption (by for example number of hospitalisations, length of stay, specialist outpatient visits), types of cancer treatment or cancer staging—yet this is challenging for a cross-cancer overall analyzes.

We expected more frequent recurrences among individuals with gastrointestinal cancer because of gut microbiome disturbances, local (neo) adjuvant treatment etc., but found a lower odds; findings which have, to our knowledge, not been described in the literature, but are interesting from a mechanistic point of view, if this association is indeed true and not the result of residual confounding and/or biases. Yet, the highest, yet statistically unsignificant, odds of recurrence was found in oral and mesothelial cancer. This might suggest that anatomical location of the cancer may affect CDI infection or recurrence, beyond the immunocompromised state of the patient and/or different treatment practices including (neo-)adjuvant treatment, antibiotic prophylaxis, surgery and others. Interestingly, survivors with esophageal cancer exhibited a significantly reduced odds of CDI recurrence. Several explanations can be hypothesized, including different (neo-)adjuvant and other treatment regimens affecting the immunocompromised state of the patient and their gut and other microbiome compositions, and therefore risk of recurrence [55‐58]. Gut biofilms may serve as a reservoir for C. difficile [59, 60], as well as other potential microbiome reservoirs such as the appendix [61, 62], resulting in relapses of the same infection, instead of a true recurrence [35, 63, 64]. As gastroesophageal reflux is a major risk factor for esophageal adenocarcinoma, increased stomach acid levels (gastric barrier function) [65], may create a challenging environment for C. difficile, and therefore hinder re-introduction of C. difficile spores—yet a large majority of individuals with oesophageal cancer is exposed to maintenance therapy with PPIs. The more frequent recurrence in oral cancer might suggest a role of the oral microbiome, and/or confounding effects of smoking and alcohol [66‐69].

Our results may have raised more questions than answers on how C. difficile strains and spores may travel through the body and interact with different microbiome niches and cancer micro-environments; and also bring back the discussion if a recurrence is due to a relapse (same strain) or a new infection (new strain) [70‐72]. Yet, we do want to stress the clinical implication that CDI is an important complication for individuals with a recent history of cancer but also for others, as mortality and of recurrence are common. There are possibilities for prevention of CDI and CDI recurrence, including an adapted, more personalized prescription regimen of antibiotics and PPIs [51, 73‐76]. Besides the additional morbidity and mortality in patients with cancer, the occurrence of CDI in these patients also conflicts with their ongoing treatment plans by reducing or delaying cancer care [25, 52].

In conclusion, our study suggests that individuals with (a history of) cancer do not necessarily have more frequent CDI recurrences compared to those without cancer, yet CDI seems an important contributing factor to mortality, and therefore the potential consequences of recurrence may be large in individuals with cancer. The risk of CDI recurrence seems to vary among different cancer types which may bring interesting insights in how CDI recurs. These findings warrant further exploration particularly in individuals with cancer along the gastro-intestinal tract, to validate or disprove our results, and to also explore the effects of cancer staging and treatment.