Introduction:

Recently, a high-resolution metabolic MRI method was introduced to map metabolic heterogeneity even within, for example, an individual breast tumor.^1-3 The use of Shutter-Speed [SS] [Dynamic-Contrast-Enhanced] DCE-MRI allows evaluation of the k_io biomarker, the rate constant for equilibrium cellular water efflux. It has been shown that k_io is proportional to the cellular metabolic rate of Na⁺,K⁺‑ATPase [^cMR_NKA], the vital cell membrane ion pump whose activity is a major goal of intermediary metabolism.¹ Interestingly, whole tumor k_io values, <k_io>_tum, averaged over sub‑populations of breast cancer subjects undergoing neoadjuvant chemotherapy [NACT] showed significant changes after just the first NACT cycle.⁴ For those who would exhibit complete [pathology-assessed] response after all cycles, <<k_io>_tum>_n decreased by 24% after the first cycle while, averaged for all other subjects, it increased.⁴ For truly personalized cancer medicine, the accurate prediction of therapy response for an individual tumor is crucial. Here, we show changes in tumor k_io maps after only one NACT cycle.

Methods:

Twenty-eight women with grade 2-3 breast cancer (in a protocol described previously^1,4) consented to 3T DCE-MRI studies, before, during, and just after the entire NACT course. The bi-lateral, fat-suppressed 3D acquisitions included 96-128 slices, and temporal resolution 14.6-20.2 s.^1,4 The nominal voxel volumes were (0.94-1.1 mm)² x 1.4 mm. The data were analyzed with an SS model assuming a single ¹H₂O signal.^1,4,5

Results:

Results from two exemplary subjects are shown in the Figure. Panels a and b display k_io maps obtained before NACT for grade 2 invasive ductal carcinoma tumors that were HER2 + [ER -, PR -] and luminal B [ER +, PR +]/HER2 + genomic molecular subtypes, respectively. The tumor k_io maps after one NACT cycle is shown in panels c and d, respectively. Since three weeks separate the acquisitions of a/c and b/d, the image slices cannot be perfectly registered. Before NACT, the two tumors differ. The luminal B tumor [b] is smaller [3.4 cc] than the HER2 tumor [a] [20.0 cc], but has a larger proportion of elevated k_io. In the luminal B image slice [b] 20%, while in the HER2 slice [a] 14%, of pixels have k_io exceeding 10 s^-1. Correspondingly, the <k_io>_tum values are 3.1 and 1.7 s^-1 for the whole luminal B and HER2 tumors, respectively. The therapy responses also differ. Both tumor volumes were reduced by the first NACT cycles: luminal B [b/d] to 1.6 cc [53% down]; and HER2 [a/c], to 14.1 cc [29% down]. Pharmacokinetic biomarkers indicate that tumor capillary density also decreases after one NACT treatment for each tumor [not shown]. However more interestingly, the k_io responses differ in sign. After first NACT, high k_io [≥10 s^-1] pixels decrease to 16% for the luminal B tumor slice [d] but actually increase to 15% for the HER2 tumor slice [c].

Discussion:

Invasive carcinoma cells have large k_io,⁶ possibly due to the membrane ion channel overexpression in cancer cells, causing chronic depolarization and altered ion gradients. This has been linked directly to the K‑Ras signaling system for uncontrolled cell proliferation.⁷ The ^cMR_NKA would be expected to be elevated, in a [futile] attempt to restore the membrane potential. After 18 weeks of NACT [six 3 week and eight 2 week cycles, respectively], the HER2 tumor was found by surgical pathology to still have ~7.5% cancer cells, while the luminal B tumor was cancer-free. Thus, the k_io responses after one NACT cycle for these individual patients reflected the group-averaged predictions for therapeutic response.⁴ The <k_io>_tum values increased from 1.7 to 2.3 s^-1 and remained unchanged [at 3.1 s^-1], respectively, for the HER2 and luminal B tumors. Even in the map image slices, the fraction of high k_io pixels increased from 14% to 15% in [a/c] while it decreased from 20% to 16% in [b/d], respectively. The HER2 patient died from her cancer four years post-diagnosis. Changing to a novel therapy trial after the first cycle k_io finding might possibly have prolonged survival. There is another interesting aspect of the a/c and b/d image pairs. In almost every instance, a k_io hot spot has its value decreased after one therapy cycle, but other regions show k_io increases. This is consistent with heterogeneous intra-tumor cellular metabolic competition; a corollary of which is that cytosidal therapy could preferentially kill the more metabolically active cell types, leaving others to flourish.⁸

Acknowledgements

Grant Support: NIH: UO1-CA154602; R44-CA180425

References

1. Springer, Li, Tudorica, Oh, Roy, Chiu, Naik, Holtorf, Afzal, Rooney, Huang, NMR Biomed. (2014) 27: 760-773. 2. Huang, Tudorica, Oh, Chiu, Roy, Troxell, Naik, Kemmer, Chen, Holtorf, Afzal, Li, Springer, PISMRM (2016) 24: 2653. 3. submitted to this meeting. 4. Tudorica, Oh, Chiu, Roy, Troxell, Naik, Kemmer, Chen, Holtorf, Afzal, Springer, Li, Huang, Trans. Oncol. (2015) 9: 8-17. 5. Li, Cai, Moloney, Chen, Huang, Woods, Coakley, Rooney, Garzotto, Springer, JMR (2016) 269: 104-112. 6. Springer, Li, Tudorica, Oh, Chui, Roy, Troxell, Naik, Kemmer, Chen, Holtorf, Afzal, Huang, PISMRM (2016) 24: 2722. 7. Zhou, Wong, Cho, van der Hoeven, Liang, Thakur, Luo, Babic, Zinsmaier, Zhu, Hu, Venkatachalam, Hancock, Science (2015) 349: 873-876. 8. Marusyk, Polyak, Science (2013) 339: 528-529.