Tracking Colorectal Organoids Separately Reveals Drug Responses Missed by Bulk Assays
Longitudinal diameter and metabolic imaging detected complementary responses in colorectal cancer organoids. Sensitive classifications often aligned with patient response, while resistant classifications still included clinical responders.
Following the Same Organoid Through Treatment
A University of Wisconsin–Madison team compared several ways of measuring treatment response in patient-derived colorectal cancer organoids. Instead of relying solely on a final image or a whole-well viability signal, the investigators followed individual organoids over time and measured their metabolic state. Their Cancers study examined what each readout detected and how selected classifications related to chemotherapy and chemoradiation outcomes.
The initial FOLFOX collection comprised 13 samples from 13 patients. A locally advanced rectal cancer collection comprised 17 samples from 13 patients, including samples from different sites in some individuals. The clinical-response analyses subsequently covered 20 chemotherapy subjects and 11 treatment-naive chemoradiation subjects. These analysis sets capture different sampling and treatment contexts.
Growth and Metabolism Reported Different Aspects of Response
The drug experiments used 10 μM 5-fluorouracil, 5 μM oxaliplatin and 1.5 nM SN38, the active metabolite of irinotecan, in specified combinations representing FOLFOX- or FOLFIRI-related treatment. Radiation experiments used 2 Gy. The Methods describe 48 hours of drug exposure followed by 48 hours without drug, with final assessment on day four.
The team compared change in an individual organoid’s diameter with its endpoint diameter, whole-well ATP signal and optical redox ratio. The latter uses NAD(P)H and FAD autofluorescence to assess a metabolic response without destroying the organoid. Tracking change within the same structure reduces the influence of its starting size, while whole-well assays combine the signals from all structures and cells present.
Individual diameter change often detected responses missed by a bulk or single-time-point measurement, but the assays were complementary. The Results text reports that, for one rectal cancer model, endpoint size and redox measurements detected significant effects that longitudinal diameter and ATP did not. In another, diameter change detected an effect that the other measurements missed. The study therefore supports combining biological readouts rather than assuming that one measurement captures every response.
Clinical Concordance Was Stronger for Sensitive Classifications
Among the 20 chemotherapy subjects, all seven whose organoids were classified as sensitive had a clinical response. Response occurred in 57% of the seven intermediate cases and 33% of the six resistant cases. These categories provided a gradient in observed response, but the resistant group contained patients who benefited clinically.
The chemoradiation analysis showed a similar asymmetry. All five sensitive cases responded, including two complete responses. Two of three intermediate cases and two of three resistant cases also had partial responses. A resistant culture result consequently had limited ability to identify patients who would fail to respond.
Combining diameter and redox measurements identified a sensitive group of 12 subjects, of whom 92% had a partial or complete response; the remaining subject had stable disease, and none had progressive disease. Half of the subjects classified as resistant by the combined analysis nevertheless had a partial response. A paired diagnosis/recurrence example also showed that organoid responses could be assessed at different treatment stages, although a single longitudinal clinical example cannot establish predictive performance.
Model Evolution and Threshold Choice Affect Interpretation
Genomic comparisons found that the cultures retained 76% of driver mutations identified in the source material. Additional driver alterations were detected in 57% of samples, and unique subclonal mutations were present in 50%. Thus, substantial tumor information remained alongside evidence of sampling differences or evolution during culture.
The chemotherapy classification thresholds drew on earlier work, whereas thresholds for radiation were chosen within this study. The authors also note that an earlier redox measurement may capture some metabolic responses better than the day-four assessment. Independent evaluation of timing and classification is needed before these readouts can guide treatment choices. The current evidence is strongest for exposing response heterogeneity and showing why assay choice changes the result.