Which Reference Organisms Exercise Which Rapid Panel Targets? An Open Crosswalk of the CDC & FDA Antimicrobial Resistance Isolate Bank to FDA-Cleared Blood Culture Identification Panels, with an External Reference-Method Check of a U.S. Benchmark

Before a laboratory reports results from a new test, U.S. federal rules require it to verify the test’s performance, and characterized reference organisms are the usual material. The CDC & FDA Antimicrobial Resistance (AR) Isolate Bank is the main free U.S. source of such organisms, but it is organized by panel rather than by test, so a laboratory introducing a rapid blood culture identification panel cannot easily see which isolates exercise which resistance target. We built an open, versioned crosswalk from the Bank’s public pages. For all 1,087 isolates in 39 panels we recorded the CDC gene list, the BioSample accession (975 isolates) and CDC’s reference broth microdilution results. We mapped 578 distinct CDC gene names to targets through the NCBI AMRFinderPlus Reference Gene Hierarchy and aligned them with the resistance targets of 7 FDA-cleared panels, taken from 9 FDA decision summaries. Of 825 isolates in organisms these panels test, 529 (64.1%) carry at least one target. Every reported target except OXA-143-like is represented, from CTX-M (168 isolates in 17 panels) to mecC (2). Every panel’s targets can be covered by two to five Bank isolates. We flag 169 isolates whose gene and growth results part ways. Among them are 14 that carry a target while testing not resistant (KPC with meropenem MICs of 0.25–2 mg/L, mecA-positive S. aureus susceptible to oxacillin and cefoxitin, a vanA operon lacking its regulators), and 38 carbapenem-resistant Enterobacterales and Acinetobacter without any panel carbapenemase. We then used the Bank as an external yardstick for a published U.S. genotype–phenotype benchmark. NCBI AMRFinderPlus reproduced CDC’s target calls (Cohen’s κ = 1.00 for eight targets, 0.93 for OXA-48-like, 0.98 for CTX-M). The benchmark’s breakpoint rules reproduced CDC’s interpretations for 99.8% of 2,851 results. In 538 independent isolates, the benchmark’s gene-level estimates held wherever both could be estimated: a panel target predicted resistance to at least one carbapenem with PPV 96.6%, and CTX-M predicted ceftriaxone resistance with PPV 100.0%. The crosswalk, code, provenance log and quality-control report are openly released.