*Helicobacter pylori* (*H. pylori*) infects over 50% of the global population, often asymptomatically, but can lead to serious complications like chronic gastritis, peptic ulcers, and gastric cancer. Clarithromycin, a macrolide antibiotic, is crucial in *H. pylori* eradication treatments, but resistance is a growing problem leading to treatment failures and increased prevalence of the infection and its associated diseases. Clarithromycin resistance (Cla-res) primarily arises from point mutations in the 23S rRNA gene, affecting the antibiotic's binding to the bacterial ribosome. Efflux pumps may also contribute to resistance, but mainly in conjunction with these mutations. While horizontal gene transfer can spread resistance within an individual, it's not believed to transmit resistance between individuals. The increasing Cla-res prevalence reduces the efficacy of standard triple therapies, making Cla-res *H. pylori* a significant global health concern. International guidelines recommend susceptibility testing before using clarithromycin-containing regimens where Cla-res rates exceed 15%. Previous studies showed cross-resistance among macrolides and a link between macrolide use and Cla-res. Primary Cla-res, usually defined as resistance in treatment-naive patients, might also occur due to clarithromycin or other macrolide use for non-eradication purposes. The role of non-eradication macrolide therapies in maintaining Cla-res prevalence is unclear, and epidemiological modeling can help clarify this. This study aimed to determine the true primary Cla-res rate by analyzing prior macrolide use in a large cohort and establishing the role of macrolides (particularly non-eradication use) in high Cla-res rates. A mathematical model was used to analyze the dynamics of Cla-res *H. pylori* infections and identify the source of primary resistance.

Existing literature confirms the crucial role of clarithromycin in *H. pylori* eradication and the significant impact of antibiotic resistance on treatment success. Studies have demonstrated the mechanisms of clarithromycin resistance, focusing on point mutations in the 23S rRNA gene and the synergistic role of efflux pumps. The limited contribution of horizontal gene transfer to inter-individual transmission of resistance is also well-documented. The increased prevalence of Cla-res strains has been highlighted in numerous studies, alongside the declining efficacy of empirical triple therapy and the urgent need for susceptibility testing in high-resistance settings. Previous research has established a clear correlation between macrolide consumption and the development of Cla-res in *H. pylori*, although defining 'primary' resistance has varied, often relying on incomplete patient history data or population-level antibiotic sales data. The precise impact of non-eradication macrolide use on both primary and secondary resistance remains incompletely understood. Prior epidemiological modeling of *H. pylori* dynamics has primarily focused on infection transmission without explicit incorporation of antibiotic resistance, highlighting a gap in the understanding of the population-level impact of resistance.

This retrospective observational study involved 4744 *H. pylori*-infected patients from Central Hungary. Clarithromycin susceptibility was determined using fluorescence in situ hybridization (FISH) on fixed gastric tissue samples. FISH detected *H. pylori* using a species-specific probe and probes targeting the three most common Cla-res mutations (A2143G, A2144G, A2143C) in the 23S rRNA gene. Bacterial density was categorized as 1+, 2+, or 3+. Patient macrolide use (including clarithromycin and other macrolides) was determined using a national medicine-dispensing database, differentiating between macrolide use for *H. pylori* eradication and other purposes. Primary resistance was defined as resistance in patients with no prior macrolide consumption. Statistical analyses, including Fisher’s exact test and generalized linear models, were used to assess the association between resistance, age, sex, bacterial density, and macrolide use. A compartmental mathematical model was developed to analyze the population dynamics of Cla-res *H. pylori*, considering the transmission of susceptible and resistant strains, spontaneous mutations, the effect of macrolide use (both for eradication and other purposes), and treatment outcomes. The model incorporated parameters such as demographic turnover, rates of macrolide use and eradication treatments, spontaneous mutation rates, and the relative transmission fitness of resistant strains. The model was calibrated using data from the cohort and published literature, and simulations were performed to predict the future prevalence of Cla-res under different scenarios (continued current macrolide use vs. discontinuation of non-eradication use). The model incorporated 12 compartments representing various infection and medication history states (uninfected, wild-type infected, resistant infected, heteroresistant, each further categorized by macrolide history: macrolide-naive, macrolide-exposed, clarithromycin-exposed).

The study found an overall Cla-res prevalence of 17.2% (816/4744). Females showed a significantly higher Cla-res rate (19.8%) than males (13.7%). Heteroresistance was detected in 47.2% of resistant cases. The conventional definition of primary Cla-res (no prior clarithromycin-containing eradication treatment) yielded a rate of 13.3%, with a significant sex difference. However, a more accurate definition excluding all prior macrolide use revealed a significantly lower primary Cla-res rate of 5.5%, with no significant sex difference. Secondary resistance (in macrolide-exposed patients) was 30.6%, significantly higher than primary resistance. Women showed a significantly higher secondary resistance rate than men, with higher average macrolide dispensing occasions. The mathematical model predicted that 98.7% of macrolide-naive Cla-res infections originated from transmission of resistant bacteria, with only 1.3% from spontaneous mutations. The relative transmission fitness of resistant *H. pylori* was estimated to be 0.72. Discontinuing macrolide use for non-eradication purposes would significantly reduce the growth rate of clarithromycin resistance, although not eliminate it entirely, as already existing cases remain resistant and transmission of resistance continues.

This study provides a comprehensive analysis of clarithromycin resistance in *H. pylori*, utilizing a large cohort and a novel mathematical model. The findings highlight the importance of considering all prior macrolide use when assessing primary resistance, revealing a substantially lower true primary resistance rate than previously thought. The dominant role of transmission of resistant strains in the spread of primary resistance underscores the significance of community-level control strategies. The model predicts a beneficial impact of restricting macrolide use for non-eradication purposes in slowing the increase in Cla-res prevalence. The high prevalence of heteroresistance detected in this study warrants further research to optimize treatment strategies for these complex cases. The study supports the need for susceptibility testing prior to clarithromycin-containing treatment, particularly in high-risk subgroups identified by sex and age, and confirms that unsuccessful eradication attempts represent a major risk factor for the development of secondary resistance.

This study demonstrates the importance of a refined definition of primary clarithromycin resistance in *H. pylori* and the crucial role of transmission of resistant strains in its spread. The mathematical model provides a valuable tool for predicting future resistance trends and informing public health interventions. Minimizing non-eradication macrolide use is crucial in slowing the rise of resistance. Further research should focus on optimizing therapeutic strategies for heteroresistant infections and developing novel eradication regimens to combat resistance.

The FISH assay only detected the three most common Cla-res mutations, potentially underestimating resistance caused by other mechanisms. The mathematical model involved some simplifying assumptions, such as homogeneous mixing and mass-action incidence, and assumed a representative sample of macrolide use for non-eradication purposes, while not assuming this for the eradication treatment rate, as the cohort represents *H. pylori*-infected patients who are more likely to receive these treatments. The model also lacked detailed data on the homo-/heteroresistance ratio for each subgroup.