Antimicrobial resistance (AMR) and drug-resistant tuberculosis (DR TB) are eroding the effectiveness of routine care, particularly in low- and middle-income settings. We propose a practical, two phase wastewater based surveillance platform in Lima that detects genes conferring resistance to selected antibiotics among Gram negative pathogens and resistance markers in Mycobacterium tuberculosis. The goal is to identify a low cost workflow that tracks real world resistance patterns with sufficient speed, sensitivity, and interpretability to inform stewardship and early warning.

Where we will work and what we will sample. We will monitor three types of wastewater: (i) hospital sewer outlets at Hospital María Auxiliadora (south Lima), Hospital Cayetano Heredia (north Lima), and Hospital del Niño (central Lima); (ii) a community sewer catchment in the district of Villa El Salvador (south Lima); and (iii) pooled drains from an animal slaughter/sales network comprising a poultry market and the Yerbateros municipal abattoir (cattle, swine, and sheep). Partner hospitals will provide aggregate, de-identified data aligned to the study period. Specifically, we will receive: (i) monthly counts of bacterial isolates from routine clinical microbiology (e.g., E. coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Acinetobacter baumannii); (ii) for each isolate, the proportion resistant to sentinel antibiotics of interest (ceftazidime, ciprofloxacin, meropenem), as reported in their antibiograms; (iii) denominator data on the total number of cultures performed per specimen type (urine, blood, respiratory, wound); and (iv) aggregate tuberculosis indicators, including the number of notified TB cases, the proportion positive on Xpert MTB/RIF, and among positives, the proportions resistant to rifampicin, isoniazid, and pyrazinamide (where available). Epidemiological data from the community will be obtained from aggregated health center records in the catchment area. All data will be provided as monthly time-series, anonymized with no individual-level identifiers, and used solely to compare temporal coherence with wastewater-derived signals.

How we will compare methods (Phase 1, Year 1). In Phase 1 (Year 1), using wastewater from hospital outlets (with hospital resistance data as a benchmark), we will conduct a single head-to-head experiment comparing three candidate procedures in parallel on the same samples. The goal is to determine which procedure performs best for detecting antibiotic-resistant Gram-negative bacteria so it can be adopted in Phase 2. Each procedure begins with selective antibiotic broth enrichment to recover resistant organisms, but they diverge in the detection workflow:

-P1: Plate the enriched broth to isolate bacterial colonies, identify species by MALDI-TOF mass spectrometry, and perform whole-genome sequencing (WGS) on a non-redundant set of isolates using Oxford Nanopore Technologies (ONT).
-P2: Skip culturing and perform direct metagenomic sequencing (ONT) on the enriched broth to capture the resistome (the collection of resistance genes present in the sample).
-P3: Apply rapid lateral-flow immunochromatographic tests (LFT) to the enriched broth to detect specific resistance enzymes (e.g., β-lactamases) via antigen capture.

We will evaluate all three approaches side-by-side using predefined performance metrics such as analytical sensitivity/specificity, reproducibility, concordance with short-lag hospital resistance time-series, time-to-result, per-sample cost, and ease of implementation. This comparison will focus on three critical antibiotic classes, a third-generation cephalosporin (ceftazidime), a fluoroquinolone (ciprofloxacin), and a carbapenem (meropenem), and their key resistance genes (for example, extended-spectrum β-lactamase genes like blaCTX-M, plasmid-mediated quinolone resistance genes like qnr, and carbapenemase genes such as blaNDM, blaKPC, or blaOXA-48-like). Detected gene signals will be mapped to their corresponding phenotypic resistance and interpreted in a clinical context (for instance, relating results to newer combination therapies such as ceftazidime-avibactam). By the end of Year 1, this integrated comparison will reveal the best-performing procedure, which we will carry forward for Phase 2 implementation.

How we will extend to the city (Phase 2, Years 2-3). After selecting the “best procedure” from Phase 1, we will deploy it to the Villa El Salvador community catchment and to the pooled poultry-market + Yerbateros abattoir drains to measure trends and build a community level antibiogram. We will link wastewater indicators to de identified community clinical aggregates (obtained from health center statistics) and develop statistical models to assess whether changes in environmental gene signals precede, and help predict, changes in clinical resistance.
TB early warning, ethics, and communication. In parallel, we will screen wastewater DNA for M. tuberculosis markers (IS6110 and known drug-resistance mutations in rpoB, katG, and pncA) as a qualitative early-warning signal (from the community and hospital sources only); we will not culture TB from environmental matrices. All clinical linkages will use aggregate, anonymized data and adhere to strict data governance protocols. Any sensitive findings (e.g., persistent high-risk gene signals at a given site) will be communicated discreetly to institutional partners. We will evaluate how well the wastewater findings correlate with the community and hospital burden of TB (including MDR-TB and PZA-resistant TB).

Expected products. By the end of the project we expect: (i) a validated, low cost workflow for wastewater AMR surveillance in Gram-negative bacteria and M. tuberculosis; (ii) matrix specific, decision ready antibiograms for hospital, community, and animal processing drains; and (iii) an initial early warning model that integrates environmental and clinical trends to support timely stewardship and public health action in Lima.To maximize public-health impact and align with Templeton’s emphasis on population benefit, we will not stop at method development. From the outset, we will work with the Peruvian Ministry of Health (MINSA), the National Institute of Health (INS), SEDAPAL (the Lima water and sanitation utility), municipal and regional health authorities, and hospital infection-prevention and stewardship teams to co-design how this workflow could be embedded in routine practice. We anticipate a stepwise pathway in which our team initially provides centralized testing and analytical support while local laboratories and surveillance units are trained, followed by gradual transfer of sampling and analysis to municipal or hospital laboratories within existing surveillance networks. The study’s dashboard, protocols, and costed workflow descriptions will be used in joint planning sessions to define realistic options for city-wide and, eventually, regional deployment so that wastewater-based AMR/TB surveillance can directly inform stewardship and public-health action at the population level.

In parallel with these core aims, we will conduct a small, exploratory proof-of-concept study to adapt an electrochemical CRISPR-Cas12 biosensor platform, previously piloted by our team in TB diagnostics, for two sentinel gene targets in wastewater: IS6110 (M. tuberculosis), representing a low-abundance, high-priority pathogen, and blaCTX-M, representing a high-abundance ESBL marker in Gram-negative bacteria. Using a subset of wastewater concentrates already processed by qPCR and sequencing, we will compare biosensor readouts against laboratory benchmarks to estimate analytical performance and operational feasibility. This exploratory component is not required to achieve the main aims of the project, but it will generate critical preliminary data on whether a simple, low-cost, field-adaptable device could realistically support the future scaling of wastewater-based AMR/TB surveillance in Peru.