The present study sequentially analyzed the dynamic changes of IVIM-parametric coupled with the histopathological features in different radio-sensitive NPC xenografts underwent the fractional radiations. After radiations, CNE-2 xenografts of higher radiosensitivity presented greater changes on the IVIM-parametric than lower radiosensitive CNE-1 xenografts, as well as the histopathological features. Parametric D correlated negatively with cellularity and positively with necrotic proportion of xenografts, whereas, f correlated positively with cellularity and negatively with necrosis proportion. It is suggested sequential IVIM-parametric could provide valuable bio-information in characterizing the histopathological features of NPC xenografts.
To evaluate the heterogeneity of diffusion in vivo effectively, Le Behan D et al. [5, 6] proposed a theory of IVIM basing on a bi-exponential model. As for IVIM-parametric, D represents the true diffusion of water molecular in extracellular space (i.e. slow diffusion pool), D* characterizes the perfusion-related diffusion of microcirculation in the capillary networks (i.e. fast diffusion pool) and f is the fraction of perfusion-related diffusion, respectively. Thus, the semi-quantitative IVIM-parametric could further clarify and distinguish the contribution of fast diffusion pool from that of slow diffusion pool basing on the bi-exponential model of IVIM theory[16, 17]. In this study, we observed a transient decline on D parametric after the fractional radiation of 10Gy, which then increased continuously. However, the cellularity and necrosis proportion of xenografts didn’t change significantly until 20Gy radiations. The potential reason might be that early in the course of irradiation, the irradiation-related cell swelling was of common phenomenon among tumor tissues. Before the process of necrocytosis or apoptosis, the expression of Na+-K+-ATPase on the membrane firstly decreased dramatically due to the hypoxia condition in tumor tissues. Even that the integrity of membrane was not affected, the trans-membrane transport of water molecule has declined markedly, leading to the swelling of tumor cells and a narrower extracellular space, and making the aggravated restriction of water diffusion in extra-cellular space. Pan J et al. had reported a obvious decline on ADC value of NPC xenografts early after a single fraction of radiations, which is consistent with our present results.
As the process of fractional radiation moving forward, the Caspase 3 would be activated to further motivate the process of cell apoptosis due to the accumulation of irradiation dose [19, 20]. Therefore, the cellularity of tumor tissues declined while the necrosis areas increased much more obviously, and the freedom of water diffusion was less restricted, resulting in a notable increase on parametric D. On the other hand, radiations cause the swelling of vascular endothelial cell in the capillary networks as well, resulting in a more serious microcirculation disturbance and down-regulating the perfusion level in tumor tissues. Accordingly, D* and f of xenografts exhibited a significant decline after fractional radiations, in particular, the later stage of irradiation. In addition, the corresponding changes of parametric D, f, and D* after radiations were found more greater in CNE-2 xenografts of higher radiosensitivity relative to the lower radiosensitive CNE-1 xenografts. These above observations suggested dynamic changes of IVIM-parametric could indirectly characterize the microstructures and radiosensitivity, and potentially provide a bio-information on tumor tissues of NPC xenografts receiving the fractional radiations.
Previous studies have demonstrated IVIM-parametric not only useful for characterizing the diffusion and perfusion features of malignancies, but also correlated well with patient’s treatment response to chemotherapy and/or radiotherapy [10–13]. Our preliminary results indicated IVIM-parametric could effectively reflect the radiosensitivity of NPC xenografts and potentially clarify their dynamic changes of microstructures after fractional radiations. D correlated negatively with cellularity and positively with necrosis proportion, and f exhibited a positive correlation with cellularity and a negative correlation with necrosis proportion. Similarly, Puig J et al.  contended IVIM-parametric were useful for characterizing the diffusion and perfusion features of glioblastoma, in which they observed a significant positive correlation between f and microvessel density as well as a negative correlation between cellularity and D. As confirmed by histology, parametric f and D* were associated siginificantly with the angiogenesis degree of microvessels and determined upon the attenuation of microvascular, in particular, the basement membrane thickness coupled with pericyte coverage. Thus, D* parametric was accounted as a proportional correlation with the average blood velocity as well as capillary segment length [22–24]. These founding indirectly revealed that regions with the highest cellularity in tumor tissues always correlated with the maximum vascularity area, and maps of IVIM-parametric could also highlight tumor’s heterogeneous pathological features generally.
Nevertheless, importantly, it is consistent with previous reports that a relatively higher coefficient variations (CVs) were observed on parametric D* and f, ranging from 12.78–24.12%. Lai V et al. [7, 25] also believed the higher variability in the measurement of parametric D* and f lay heavily on the method of drawing ROI manually. Therefore, in this present study, two radiologists conducted the analysis and calculation of IVIM-parametric with a double-blind control to reduce the variability and error of measurements. Besides, two observers initially reached an agreement on the edge definition of target lesions before manually contouring the ROI. It was suggested the ROI should be delineated along the edge so far as possible to include the whole tumor of xenografts, paying attentions to avoid the interference of the intra-lesion necrosis, adjacent bone, air, and other structures, etc. The inter- and intra-observer ICCs for parametric D, D* and f were observed ranging from 0.851 to 0.973, indicating a excellent reproducibility and consistency on the measurements of IVIM-parametric between observers in this present study.
This study still has several limitations. Firstly, the IVIM-parametric and pathological features of xenografts were derived and originated from different irradiation groups, and there always existed a certain degree of heterogeneity between xenograft’s tissues themselves. Another limitation is that the ROIs placed on images of IVIM-parametric were drawn manually including the whole tumor of xenografts, whereas, the pathological features analyzed on HE images were selected randomly which is partly representative and could not reflect the heterogeneity of integrated tumor tissues as the IVIM sections. Moreover, the observation period is not enough to characterize the xenograft’s micro-environment since the dynamic changes of IVIM-parametric and pathological features might continue for such an extended period after radiations.