Although recurrent cancers are often aggressive, little is known about the intracellular events required for cancer recurrences. Due to this lack of mechanistic information, there is no test to predict cancer recurrences or non-recurrences during early stages of disease. In this retrospective study, we use ductal carcinoma in situ (DCIS) of the breast as a framework to better understand the mechanism of cancer recurrences using patient outcomes as the physiological observable. Conventional pathology slides were labeled with anti-phosphofructokinase type L (PFKL) and anti-phosphofructokinase/fructose-2,6-bisphosphatase type 4 (PFKFB4) reagents. PFKL and PFKFB4 were found in ductal epithelial cell nucleoli from DCIS samples of women who did not experience a cancer recurrence. In contrast, PFKL and PFKFB4 may be found near the plasma membrane in samples from patients who will develop recurrent cancer. Using machine learning to predict patient outcomes, holdout studies of individual patient micrographs for the three biomarkers PFKL, PFKFB4, and phosphorylated GLUT1 demonstrated 38.6% true negatives, 49.5% true positives, 11.9% false positives and 0% false negatives (N=101). A sub-population of recurrent samples demonstrated PFKL, PFKFB4, and phosphorylated glucose transporter 1 accumulation at the apical surface of epithelial cells, suggesting that carbohydrates can be harvested from the ducts' luminal spaces as an energy source. We suggest that PFK isotype patterns are metabolic switches representing key mechanistic steps of recurrences. Furthermore, PFK enzyme patterns within epithelial cells contribute to an accurate diagnostic test to classify DCIS patients as high or low recurrence risk.

Although the existence of non-recurrent and recurrent forms of ductal carcinoma in situ (DCIS) of the breast are observed, no evidence-based test can make this distinction. This retrospective case-control study used archival DCIS samples stained with anti-phospho-Ser226-GLUT1 (glucose transporter type 1) and anti-phosphofructokinase type L (PFKL) antibodies. Immunofluorescence micrographs were used to create machine learning (ML) models of recurrent and non-recurrent biomarker patterns, which were evaluated in cross-validation studies. Clinical performance was assessed by holdout studies using patients whose data were not used in training. Micrographs were stratified by the recurrence probability of each image. Recurrent patients were defined by at least one image with a probability of recurrence >98% whereas non-recurrent patients had none. These studies demonstrated no false negatives, identified true positives, and uniquely identified true negatives. Roughly 20% of the microscope fields of recurrent lesions were computationally recurrent. Strong prognostic results were obtained for both Caucasian and African American women. Our machine tool provides the first means to accurately predict recurrent and non-recurrent patient outcomes. We suggest that at least some false positives were true positives that benefitted from surgical intervention. The intracellular locations of phospho-Ser226-GLUT1 and phosphofructokinase type L likely participate in cancer recurrences by accelerating glucose flux, a key feature of the Warburg Effect.

No abstract