The coexistence of both multidrug resistance genes and virulence determinants within the same bacterial background represents a major evolutionary shift in hospital-associated K. pneumoniae. Globally, the emergence of these resistance–virulence convergent high-risk clones has largely been explained through two classical routes (36). The first is acquisition of MDR plasmids by highly virulent ST23-like lineages. The second is transfer of pLVPK-like virulence plasmids into established multidrug-resistant clones, particularly ST11 (37, 38). The latter mechanism has been reported most frequently in South and East Asia where it has driven several outbreaks of carbapenem-resistant hypervirulent K. pneumoniae (27, 41). However, resistance–virulence convergence is an ongoing and complex evolutionary process, and different lineages may follow distinct pathways to reach similar convergent states. Our findings support the view that acquisition and long-term maintenance of complete virulence plasmids by multidrug-resistant clones may be relatively uncommon. Classical pLVPK-like virulence plasmids are largely restricted to specific hypervirulent genetic backgrounds and appear difficult to disseminate widely among multidrug-resistant lineages because they are typically non-self-transmissible and may impose substantial fitness costs in unrelated hosts (39). Consequently, successful convergence often depends on additional evolutionary mechanisms that facilitate the transfer, maintenance, and stabilization of virulence determinants within resistant lineages (40). In this regard, recent studies have highlighted the importance of homologous recombination, helper plasmids, and formation of resistance–virulence hybrid plasmids in driving convergence. These observations emphasize that the evolutionary routes leading to MDR-hvKp are considerably more diverse than initially predicted (33).

Against this background, our findings suggest that dominant Indian hospital-associated clones have adopted a distinct convergence strategy. Rather than acquiring complete virulence plasmids, these lineages selectively retain specific virulence determinants through diverse resistance–virulence plasmid architectures. Across ST147, ST2096, and ST231, the aerobactin locus was consistently present, whereas other canonical hypervirulence-associated determinants including iro, rmpA, and rmpA2 showed variable distribution or were absent altogether, defining a shared intermediate convergence state across all three lineages (30, 51). The consistent retention of aerobactin across three independent high-risk lineages suggests that it provides a selective advantage. As a major virulence determinant, aerobactin plays a key role in iron acquisition, extraintestinal dissemination, and host adaptation (35, 49, 51). In contrast, maintenance of complete virulence plasmids and highly mucoviscous phenotypes, may impose considerable metabolic costs particularly when combined with large multidrug-resistance plasmids (39). Consistent with this hypothesis, convergent CR-hvKP lineages frequently show attenuation of hypermucoviscosity and reduced expression of rmpA/rmpA2-associated virulence pathways (48,50). Consequently, convergent multidrug-resistant isolates carrying canonical virulence loci often exhibit lower-than-expected virulence phenotypes (1). Together, these observations support the view that the intermediate convergence state observed among dominant Indian hospital clones is consistent with a hospital-adapted evolutionary optimum, in which selective retention of high-value determinants such as aerobactin balances invasive potential with long-term persistence under sustained antimicrobial selection.

Within this framework, ST147 appears to represent the most dynamic route towards the shared intermediate convergence state. As one of the most globally disseminated high-risk MDR clones, ST147 has repeatedly demonstrated remarkable genomic plasticity, with a well-recognized capacity to acquire carbapenemases, ESBL determinants, and diverse mobile genetic elements in hospital settings (8, 19). Increasing evidence suggests that this success is closely linked to the ability of ST147 to repeatedly recruit and remodel accessory genetic elements, particularly IncFIB - IncHI1B hybrid plasmids carrying combinations of resistance and virulence determinants (5, 6, 21). Notably, these hybrid plasmids do not represent a single conserved virulence–resistance platform. Instead, studies from Europe and East Asia have revealed multiple related but structurally distinct IncHI1B–IncFIB backbones carrying different assortments of virulence loci, resistance genes, and mobile genetic elements (6, 52). The extensive rearrangement of insertion sequences, transposons, and resistance islands reported within ST147 populations further suggests that these plasmids are subject to continual recombination and remodeling rather than stable vertical inheritance (14). Recent evidence of ongoing in-host evolution and convergence within ST147 further reinforces the notion that this lineage remains evolutionarily active and highly adaptable (17). Consistent with this model, virulence determinants in our collection were confined to specific phylogenetic subclusters and associated with diverse IncHI1B–IncFIB hybrid plasmid architectures. Together, these observations support the view that ST147 functions as an opportunistic assembler of resistance and virulence modules, in which continual plasmid restructuring serves as the primary mechanism driving resistance–virulence convergence.

By contrast, ST2096 represents a fundamentally different convergence trajectory characterized by stabilization rather than continual remodeling. First documented in India as a multidrug-resistant hypervirulent lineage, ST2096 is consistently associated with highly conserved IncFIB–IncHI1B hybrid plasmids carrying aerobactin and rmpA2 (27). Unlike the mosaic, repeatedly remodelled plasmid configurations observed in ST147, the ST2096 hybrid appears to function as a stable, integrated unit. The presence of virulence determinants, within basal phylogenetic nodes suggests an early convergence event followed by clonal expansion (27). Similar convergent ST2096 isolates have been reported from geographically distinct settings, supporting the notion that this resistance–virulence configuration is both stable and epidemiologically successful (15, 42, 43). This pattern aligns with emerging models showing that certain IncHI1B-based hybrid plasmids can achieve stable transmission within permissive genomic backgrounds (46, 47). Notably, carbapenemase genes in this cohort were carried on separate, smaller plasmids rather than integrated into the large hybrid structure, suggesting modular coexistence of both traits without full plasmid fusion (27). ST2096 therefore appears to represent a consolidator lineage, in which a favorable resistance–virulence configuration has been evolutionarily stabilized and subsequently amplified through clonal expansion.

A third convergence strategy is exemplified by ST231, where resistance and virulence traits coexist without the formation of a large hybrid resistance–virulence plasmid. ST231 is a major carbapenem-resistant lineage in India and is strongly associated with bla_OXA-48-like, frequently carried on small ColKP3 plasmids (34, 44). Despite this conserved resistance background, virulence determinants in our collection were primarily associated with smaller IncFIA plasmids carrying the aerobactin locus. In contrast to ST147 and ST2096, large IncFIB–IncHI1B hybrid plasmids were absent, indicating that convergence can occur without physical integration of resistance and virulence determinants into a single plasmid backbone. The continued recovery of ST231 from diverse clinical and carriage settings further indicates that stable coexistence of compatible resistance and virulence plasmids can support long-term persistence and dissemination without extensive plasmid fusion (15, 45). ST231 therefore appears to represent a segregator strategy, in which resistance and virulence determinants remain physically separated while still achieving the same intermediate convergence state observed in ST147 and ST2096.

When considered together, these three lineages demonstrate that the intermediate convergence state in hospital-associated K. pneumoniae can arise through multiple evolutionary pathways. ST147 represents a highly dynamic plasmid-remodeling pathway, ST2096 a stabilized hybrid-plasmid pathway, and ST231 a segregated plasmid co-maintenance pathway. Despite these distinct trajectories, all three lineages converge on a similar biological endpoint, highlighting that clinically comparable convergent phenotypes can emerge through fundamentally different plasmid architectures. These findings also underscore the limitations of relying solely on hypermucoviscosity or single virulence markers such as aerobactin for surveillance, as isolates carrying diverse combinations of resistance and virulence determinants may lack classical hypervirulent phenotypes while still representing biologically significant intermediate forms within high-risk multidrug-resistant backgrounds. Although genomic surveillance identifying co-occurring resistance and virulence loci has substantially improved detection of convergent strains, plasmid-level resolution remains essential for understanding the acquisition, maintenance, and dissemination of these traits (31, 32). Multiplex PCR assays targeting combinations of carbapenemase genes and major virulence markers may provide an efficient first-line screening approach (4); however, lineage-specific hospital based genomic surveillance and plasmid reconstruction remain critical for tracking both clonal spread and ongoing plasmid evolution within high-risk K. pneumoniae lineages (53).

Nevertheless, several constraints in our study design must be noted. First, long-read sequencing and subsequent plasmid reconstruction were limited to representative isolates from each lineage; while this approach facilitated detailed structural comparisons, it may not capture the full spectrum of plasmid diversity within these clonal populations. Second, the study was based on a carefully selected subset of Indian clinical isolates from a single hospital rather than comprehensive sequencing of all circulating ST147, ST2096, and ST231 strains, which may limit resolution of lineage-wide evolutionary patterns. Third, functional assays including plasmid transfer kinetics, stability testing, and in vivo virulence models were not performed. Future studies integrating broader genomic surveillance with longitudinal sampling and experimental validation will be important to further define the evolutionary dynamics of resistance–virulence convergence in high-risk K. pneumoniae lineages.