Discussion

This 16-year analysis reveals an alarming AMR trajectory at our tertiary cancer center, marked by a substantial burden of MDR and XDR pathogens and predictive models forecasting further deterioration by 2030. While GNB consistently predominated, the study identifies a dual-threat landscape: rising resistance among Enterobacterales, MRSA, and VRE, contrasted by a notable decline in resistant non-fermenting GNB (NFGNB), particularly P. aeruginosa and A. baumannii-calcoaceticus which serves as a critical finding. This decline temporally coincides with institutional changes around the start of the third period, including a major facility expansion (doubled bed capacity, quadrupled clinical space) and the formal activation of an antimicrobial stewardship committee.

In patients with resistant Gram-negative bacteria (GNB), chemotherapy-induced breakdown of mucosal barriers and frequent use of invasive devices are hypothesized to provide portals of entry for enteric organisms, a mechanism supported in the literature though not directly assessed in our data set. ^{Reference Sadeghi, Rajabi and Ghanbari12} Resistant Enterobacterales is a growing concern. ESBL testing showed that more E.coli and K. pneumonia are getting resistance. This is coinciding with significantly increasing resistance to cephalosporins and carbapenems in the last period of the study (2018–2022). Carbapenem-resistant GNB phenotype is a major predictor of treatment failure and early mortality in cancer patients with systemic infections. ¹³

The proportion of S.aureus and E. faecium resistant to oxacillin and vancomycin, respectively, are increasing, possibly reflecting changes within and outside the hospital. This increase is alarming, and it is predicted that in 10 years, almost three-quarters of S.aureus isolates will be resistant to oxacillin. One area of concern is the need to rely heavily on glycopeptides or oxazolidinone, as we currently have no approved alternatives. Luckily, we did not face vancomycin resistance among any of S. aureus isolates yet. It is not easy to correlate our findings with the literature. AMR varies over time and is related to the geographic location, study period, patient population studied, age-groups, use of prophylactic antibiotics and the accuracy in reporting antimicrobial resistance. ¹⁴

Jordan has well-documented high rates of antibiotic resistance, local data highlight this burden: a study at a large public hospital reported high methicillin-resistant CoNS rates in hospitalized patients; ^{Reference Sivasankar, Goldman and Hoffman15} 46% of community-acquired UTIs in children were caused by ESBL-producing bacteria; ^{Reference Albaramki, Abdelghani and Dalaeen16} ESBL-producing E. coli was identified in 62% of UTI specimens across two tertiary hospitals; ¹⁷ and 48% of environmental E. coli isolates from five tertiary hospitals were MDR, with 68% phenotypically identified as ESBL producers. ¹⁸ MRSA prevalence is also notably high in Jordan, with a study across two tertiary hospitals finding that 73% of S. aureus clinical isolates harbored the mecA gene, confirming MRSA designation. ¹⁹

Regarding multidrug-resistant (MDR) Acinetobacter baumannii-calcoaceticus complex isolates, a report from the largest military hospital in Jordan found that more than three-quarters of isolates were highly resistant to all antibiotics, including tigecycline, with only minocycline and colistin, the last-resort agents, retaining meaningful activity (resistance rates of 27% and 1.7%, respectively). ^{Reference Batarseh, Al-Sarhan, Maayteh, Al-Khatirei and Alarmouti20} A recent Jordanian study evaluated the resistance to newly introduced cephalosporines mainly ceftazidime-avibactam and ceftolozane-tazobactam, regrettably, found that some isolates had already developed resistance to these agents prior to their widespread use in the country. ^{Reference Said and Kwaik21}

Rates of resistance seem to be higher in cancer wards. Praphash et al reported alarmingly high rates of resistance among adult cancer patients. In their study, half (49%) of all GNB were resistant to piperacellin/tazobactam and 28% were resistant to meropenem. Sensitivity to amikacin ranged from 24% to 60% and meropenem was reliable only in E. coli and K. pneumoniae (>90% sensitivity). Meropenem sensitivity was reported in only 66% and 39% of pseudomonas and Acinetobacter isolates, respectively. Interestingly, MRSA was reported only in 18% of their isolates. ^{Reference Prabhash, Medhekar and Ghadyalpatil22}

The SENTRY antimicrobial surveillance program, tracking resistance across hundreds of hospitals worldwide since 1997, reported a decline in MRSA prevalence in Western centers to approximately 39% (2013–2016) following a broader post2008 reduction in oxacillin resistance among S. aureus. ^{Reference Diekema, Pfaller, Shortridge, Zervos and Jones23} This stands in stark contrast to our center, where MRSA prevalence continues to rise. ^{Reference Diekema, Pfaller, Shortridge, Zervos and Jones23} This discrepancy suggests that regional factors, such as higher community colonization rates in Jordan or differences in decolonization practices, may be driving the institutional trend.

In contrast, the prevalence of MDR Enterobacterales increased from 6.2% in 1997 to 2000 to 15.8% in 2013 to 2016. MDR rates were highest among non-fermentative Gram-negative bacilli (GNB), and colistin was the only agent with predictable activity against Acinetobacter baumannii-Acinetobacter calcoaceticus complex (97% susceptible). ^{Reference Diekema, Hsueh and Mendes24} ESBL rates among our patients were double those reported by SENTRY, which documented a peak MDR Enterobacterales prevalence of 15.8%, far below the 68.8% recorded in our study. Notably, SENTRY reported stable VRE prevalence post2012 (16.4% overall), which remains higher than our latest VRE rate of 13%. ^{Reference Diekema, Hsueh and Mendes24}

Patients with hematologic malignancies showed a more severe, unadjusted AMR profile than those with solid tumors, with leukemia and lymphoma patients exhibiting the highest XDR rates (5.6%–7.9%) versus 4.0% in carcinoma patients, an association, not evidence of an independent diagnosis effect. VRE was of particular concern, plausibly related to prolonged neutropenia, broad-spectrum antibiotic exposure, and central-line contact common in this population, though these variables were not captured or adjusted for in our data. The finding that one-quarter of blood enterococci were vancomycin-resistant nonetheless highlights a clinically relevant gap in current prophylactic and therapeutic strategies for hematologic patients. ^{Reference Sadeghi, Rajabi and Ghanbari12}

Urine cultures showed the highest density of Gram-negative bacteria, with a 63% MDR rate among GNB isolates; the high fluoroquinolone resistance (55%) in urine-derived Enterobacterales may relate to extensive ciprofloxacin use in prophylactic and outpatient settings, though this was not directly tested. Respiratory samples, in contrast, served as a primary reservoir for resistant NFGNB and MRSA, possibly linked to mechanical ventilation and upper airway colonization in chronically ill patients, consistent with prior reports, ^{Reference Ndzabandzaba, Mothibi and Von Knorring25} though this association was not directly assessed in our data.

Our 16-year, diagnosis- and site-stratified analysis, novel linkage to the cancer registry, and NHSN/CLSI-compliant methodology are real strengths, and the predictive modeling was a genuine attempt to move beyond a static antibiogram toward something more actionable for settings like ours without real-time EMR surveillance. Still, this study has several limitations, our 2030 projections are unvalidated extrapolations of past trends assuming those trends continue, and shifts in testing platforms over 16 years may have added some residual noise despite MIC harmonization. Ongoing efforts at our center aim to develop a real-time antibiogram dashboard integrated with cancer registry data to operationalize these findings in resource-limited settings.