Microbiological and clinical characteristics of hypervirulent Klebsiella pneumoniae isolated from patients in tertiary centers: a retrospective study

Introduction

Klebsiella pneumoniae can be classified into classical (cKp) and hypervirulent (hvKp) strains based on their phenotypic and clinical characteristics. Unlike cKp, hvKp typically exhibits a hypermucoviscous phenotype, which can be detected using the string test, defined by the formation of a viscous string of at least five millimetres when a colony is stretched with an inoculation loop. In addition, hvKp has the ability to infect healthy individuals of any age and is more likely to cause infections at multiple sites and/or metastatic spread within the same host (Al Ismail et al., 2024). However, it is well recognized that hypermucoviscosity and hypervirulent are distinct yet overlapping phenotypes, as not all hypermucoviscous strains are hypervirulent, and vice versa.

Despite its hypermucoviscous nature, hvKp harbours distinct virulence determinants, such as siderophores, excessive capsule production, lipopolysaccharides, and the colibactin toxin, which enhance its survival and pathogenicity. Among these, the polysaccharide capsule is considered the key virulence factor responsible for the distinctive hypermucoviscous phenotype (Catalan-Najera, Garza-Ramos & Barrios-Camacho, 2017). Capsular serotypes K1 and K2 have been reported as the most common strains associated with hvKp infections. The ability to produce an enhanced capsular polysaccharide layer is mediated by rmpA and rmpA2 genes, which regulate the mucoid phenotype and are specific to hvKp (Choby, Howard-Anderson & Weiss, 2020; Russo & Marr, 2019). The mucoviscosity-associated gene A (magA) gene, which is recognized as a surrogate marker for K1 serotype, encodes a polymerase involved in capsule synthesis that contributes to the hypermucoviscous phenotype and enhanced virulence of K1-hvKp (Ikeda et al., 2018). Additionally, hvKp strains produce four different plasmid-encoded siderophores (aerobactin, enterobactin, salmochelin, and yersiniabactin) to facilitate iron acquisition. Among these, aerobactin, encoded by the iuc gene, is particularly significant. It is found in over 90% of hvKp strains but in only about 6% of cKp strains. Hence, aerobactin is considered a more reliable biomarker for hvKp (Kochan et al., 2023; Zhang et al., 2016).

The prevalence of hvKp varies across studies. The first clinical report published in 1986 described seven cases of invasive K. pneumoniae infection in healthy individuals from the Taiwanese community, who developed hepatic abscess in the absence of biliary tract disease, along with septic endophthalmitis (Casanova et al., 1989). The K. pneumoniae strains responsible for hepatic abscesses were more likely to exhibit a hypermucoviscous phenotype compared to non-invasive strains. Lin et al (2012) reported that healthy individuals from several Asian countries, including Malaysia, had a higher prevalence of colonic K. pneumoniae colonization (18.8% to 87.7%), compared to those from Western countries (5% to 35%) (Marr & Russo, 2019). However, data on the actual prevalence of hvKp remain limited. Therefore, this study aimed to determine the prevalence of hvKp, its associated virulence genes, and the clinical presentations of patients infected with this pathogen. In this study, hvKp is defined as K. pneumoniae isolates exhibiting a hypermucoviscous phenotype and harbouring either the KI or K2 gene. Additionally, we screened for key virulence-associated genes to complement our working definition of hvKp and to support future efforts toward more reliable, genotype-based classification, as recommended by recent studies.

Materials & Methods

Results

Clinical and microbiological characteristics of K. pneumoniae isolates

A total of 180 K. pneumoniae isolates were recovered from various types of clinical specimens across four main hospitals in Kelantan. The most common sources were tracheal aspirate (27.8%, n = 50), blood (24.4%, n = 44), and sputum (21.7%, n = 39) (Fig. 1). In line with a previous report, these findings suggest that K. pneumoniae was predominantly associated with respiratory infections in hospitalized patients (Nik Zuraina et al., 2023).

Identification of hmKp and hvKp isolates

Out of the 180 isolates, 43 (23.9%) exhibited a hypermucoviscous phenotype, as determined by a positive string test. Further molecular analysis of the capsular serotypes revealed that the prevalence of K1 and K2 was 11.1% (n = 20/180) and 6.1% (n = 11/180), respectively. The presence of these serotypes reinforces their importance as established markers in hvKp surveillance. In contrast, 82.8% (n = 149/180) of the isolates lacked both K1 and K2 capsular serotype-specific (cps) genes and were classified as non-K1/K2. Meanwhile, of the 43 hmKp isolates, 23.3% (n = 10/43) and 16.3% (n = 7/43) were K1 and K2 serotypes, respectively (Fig. 2).

To identify hvKp, K. pneumoniae isolates that exhibit a positive string test in the presence of either the K1 or K2 genes were classified as hvKp, consistent with definitions used in previous regional studies. Based on this, the prevalence of hvKp in this study was 9.4% (n = 17/180). However, it is noted that current criteria for hvKp classification increasingly emphasize the presence of specific virulence genes. Therefore, molecular screening for key virulence markers was also performed in this study to gain additional insight into the genetic characteristics of hvKp and non-hvKp isolates.

Discussion

Until today, the terminology of “hypervirulent” K. pneumoniae remains controversial and varies among studies, as there is still no consensus on a universal definition of hvKp. Even though it has been suggested that host factors, pathogen traits, and host-pathogen interactions should be considered comprehensively for defining hvKp (Liu & Guo, 2019), most studies still focus solely on the pathogen characteristics. Early reports used the hypermucoviscous phenotype to define hypervirulent strains (Lin et al., 2020; Liu et al., 2014). However, subsequent in vitro and in vivo analyses by the later studies have revealed that hypermucoviscosity alone is insufficient to reliably describe hvKp due to: (i) inconsistencies between the phenotypic hypermucoviscous and genotypic virulence characteristics (Lin et al., 2011; Zhang et al., 2015); and (ii) the limited sensitivity and sensitivity of the string test (Catalan-Najera, Garza-Ramos & Barrios-Camacho, 2017; Sanikhani et al., 2021). Hence, in parallel with hypermucoviscosity, several studies have emphasized the importance of capsular serotypes and virulence-associated genes screening, such as K1, K2, aerobactin, and rmpA genes, as more reliable characteristics to describe hvKp (Liu & Guo, 2019; Russo et al., 2014). The growing reliance on genotypic rather than phenotypic criteria for defining hvKp is further supported by recent findings that genotypic markers such as iucA and iroB have higher predictive accuracy than the hypermucoviscous phenotype (Kochan et al., 2023).

In this study, hvKp was specifically defined as K. pneumoniae strains possessing both of the following criteria: (i) hypermucoviscosity; and (ii) the presence of the K1 or K2 serotype. This approach aligns with earlier studies (Ikeda et al., 2018; Lin et al., 2020) and remains practical for preliminary identification, especially in settings with limited molecular resources. Based on these criteria, this study identified 23.9% (n = 43) hmKp and 9.4% (n = 17) hvKp among 180 K. pneumoniae isolates. Comparing our findings with previous studies, a retrospective study conducted at a single centre in Malaysia reported a lower prevalence of hmKp (7.5%) among 120 carbapenem-resistant K. pneumoniae isolates. Most of these isolates were associated with hospital-acquired or healthcare-associated infections, and none harboured K1 or K2 serotypes (Kong et al., 2021). Meanwhile, a study in China found that 33% of K. pneumoniae isolates from hospitalized patients were characterized as hmKp (Li et al., 2014). In endemic areas, the prevalence of hvKp among K. pneumoniae has been reported to range from 12% to 45% (Choby, Howard-Anderson & Weiss, 2020). A retrospective genomic epidemiology study further highlighted a higher prevalence of community-acquired hypervirulent strains in South and Southeast Asia. However, this study characterized hvKp strains solely based on genome sequencing and K. pneumoniae-specific genomic typing tools for the presence of hypervirulence-associated loci in 331 K. pneumoniae BSI isolates. The reported prevalence of these loci included K1 and K2 capsular serotypes (18%), yersiniabactin (49%), aerobactin (28%), rmpA (18%), rmpA2 (16%), and peg-344 (19%).

To further evaluate the limitations of phenotype-based hvKp identification, this study integrated virulence gene screening, revealing important discrepancies and highlighting the added value of genotypic markers for future classification. Regarding virulence factors, all the genes investigated in this study (rmpA, rmpA2, iucA, magA, and peg-344), as well as the capsular serotype antigens K1 and K2, were significantly associated with hypermucoviscosity (p < 0.05). Interestingly, the rmpA, rmpA2, iucA, and peg-344 genes were detected in all 17 hvKp isolates, which shows that these genes are strongly associated with hvKp, supporting their classification as hypervirulent based on both phenotypic and genotypic features. Among the180 isolates, the prevalence of K1 and K2 capsular serotypes was 11.1% (n = 20) and 6.1% (n = 11), respectively. These proportions were lower than those of the other tested virulence genes. Although more than 70 K-serotypes have been reported, capsular serotypes K1 and K2 are the most frequently associated with severe infections (Al Ismail et al., 2024; Liu & Guo, 2019). Additionally, K1 and K2 are the dominant capsular serotypes that are strongly related to hypervirulent strains (Liu et al., 2014; Sanikhani et al., 2021). Among them, K1 is the predominant serotype in Asia and has been reported to be highly associated with hvKp infections compared to K2 (Bilal et al., 2014). Similarly, a recent study in Malaysia found that K1 and K2 serotypes were significantly associated with hypervirulence, but not with antibiotic-resistant strains, which emphasizes the need for continuous monitoring of virulence and resistance convergence (Anuar et al., 2024).

This study also demonstrated that the associations of rmpA, rmpA2, iucA, and peg-344 virulence genes with K1 and K2 serotypes were statistically significant (p < 0.05). Aerobactin (iucA) has been used as a molecular biomarker for hvKp (Kochan et al., 2023; Russo et al., 2014). In this study, iucA was detected in 70% (n = 30) of the hmKp pathotype and was the most prevalent virulence gene among all isolates (49.4%, n = 89). However, its widespread presence suggests that relying on aerobactin as a sole biomarker for hvKp detection may not be ideal. The magA gene was significantly associated with K1 (p < 0.05), consistent with its role as a serotype-specific marker for K1 (Guo et al., 2017), justifying its absence in K2 serotypes. Although magA is widely recognized as a marker for the K1 serotype, it also plays an important role in capsule production, which contributes to the hypermucoviscous phenotype and increased virulence of K1-type hvKp strains. In this study, magA was included to confirm K1 serotype and highlight its biological role in capsule formation. Its significant association with the hypermucoviscous phenotype in our findings supports its dual relevance as both a serotype marker and a virulence-related gene specific to the K1 serotype, consistent with previous studies involving K1 hvKp isolates in Asia (Yadav, Mohanty & Behera, 2023). Based on the significant associations between these virulence genes, hypermucoviscosity, and K1/K2 capsular serotypes, the simultaneous detection of rmpA, rmpA2, magA (for K1 serotype), iucA, and peg-344 genes in a single hmKP isolate could represent a reliable genetic marker set for defining a hvKp strain in future studies. Importantly, several isolates that did not meet our hvKp definition still harbored key virulence genes, suggesting that genetically hypervirulent strains may extend beyond the classical phenotype. Future definitions may benefit from incorporating genotypic criteria to identify such putative hvKp strains, particularly when hypermucoviscosity is absent.

In addition to investigating the pathogen, this study also examined the characteristics of the host (patients) from a clinical perspective. Demographic data of the 17 cases infected by hvKp showed that more than half of the patients were under 60 years old. The finding was consistent with a previous study reporting that hvKp infections tend to affect younger, otherwise healthy individuals who present with severe disease (Marr & Russo, 2019). Specific gender was not a recognized risk factor; however, in this study, the majority of the patients infected with hvKp were male (76.5%, n = 13). The similar finding was also reported by a previous study, showing that men are more likely to be infected than women (Siu et al., 2012). In terms of clinical manifestations, most of the hvKp-infected patients were diagnosed with pneumonia (70.6%, n = 12). Cases of bacteremic community-acquired pneumonia caused by hvKp have been increasingly reported in the Asia-Pacific region and South Africa (Russo & Marr, 2019; Wu et al., 2017). Although respiratory infections are considered the second most common presentation after liver abscess (Siu et al., 2012), only one single case of liver abscess was identified in this study. This discrepancy may be attributed to geographical variations in circulating strains and exposure patterns. Hypertension was the most common underlying comorbidity (35.5%), followed by T2DM (11.8%). Some studies suggested that the association between T2DM and hvKp may vary depending on geographical location and clinical presentation (Russo & Marr, 2019). The majority of hvKp infected-patients (88%) were eventually discharged, while the mortality rate was 12% (n = 2). This was slightly lower than the mortality rate reported in previous retrospective study conducted in Beijing, China (Li et al., 2018).

One of the strengths of this study is the involvement of four major hospitals, which could provide a broader representation of the situation in Kelantan state. Additionally, the combined use of phenotypic (string test) and genotypic (multiplex PCR for virulence and capsular genes) approaches allowed for a more comprehensive characterization of hvKp. The study also evaluated clinical presentations and patient outcomes, providing clinical context to the microbiological findings. However, this study has several limitations. First, the cross-sectional design may not fully capture temporal trends or emerging clones. Second, the one-year duration of the study limited the sample collection and may have restricted the generalizability of the findings. Likewise, expanding the panel of virulence and capsular serotype genes for hvKp, such as serotypes K5, K20, K54, and K57, could improve the recovery of hypervirulent strains, although their prevalence among hvKp remains relatively low (Lin et al., 2020). Third, although AST was routinely performed as part of hospital protocols, the antimicrobial susceptibility profiles of the K. pneumoniae isolates were not included in the analysis. This limited our ability to assess potential convergence between hypervirulence and antimicrobial resistance. Given the increasing reports of multidrug-resistant hvKp strains worldwide, future surveillance studies should incorporate antibiogram data towards a better understanding of resistance trends (Chen, Zhang & Liao, 2023; Hao et al., 2020; Karampatakis, Tsergouli & Behzadi, 2023). Lastly, species-level differentiation within the K. pneumoniae complex, such as K. variicola, and K. quasipneumoniae was not performed. As conventional laboratory methods do not distinguish between these closely related taxa, it is possible that some hypervirulent strains reported here may belong to other species within the complex. Future studies should incorporate molecular tools such as whole-genome sequencing or species-specific PCR to enable more precise classification (Rodríguez-Medina et al., 2019).

Conclusions

While recent literature increasingly supports the use of genotypic markers for hvKp classification, the combined phenotypic and serotypic approach remains a practical framework, particularly in settings where molecular tools are limited. The inclusion of additional virulence gene screening in this study allowed for evaluation of the genetic characteristics underlying hypermucoviscosity and revealed the presence of hypervirulence-associated loci beyond classical hvKp definitions. These findings underscore the limitations of phenotype-based definitions and support a broader, gene-based classification in future hvKp studies. In this study, the prevalence of hvKp was 9.4%, based on the presence of both hmKp and K1 or K2 capsular serotypes. HvKp infections were more frequently observed in men and individuals with hypertension, with pneumonia being the most common clinical presentation. The mortality rate among hvKp-infected patients was 12%. Although the working definition employed was useful for identifying the traditionally defined hvKp, our findings also revealed that several non-hvKp isolates carried key virulence genes. This suggests that genetically hypervirulent strains may exist beyond phenotype-based criteria. Therefore, molecular detection of virulence factors offers a more reliable approach as the use of string test alone for hvKp detection could be misleading. These findings support the need for a more robust definition of hvKp in future studies that combines hypermucoviscosity and K1/K2 serotypes with the presence of key genotypic markers (rmpA, rmpA2, iucA, and peg-344). Adopting this combined approach may improve the accuracy of hvKp identification and enhance both clinical management and epidemiological surveillance.

Supplemental Information