The aim of this study was to evaluate imaging parameters quantifying vascularity in human cancer xenografts with two levels of vascularisation. The amount of vascularisation was determined ex vivo by a reference standard, i.e. specific histological and immunohistological methods. To provide translational results, these tumours were investigated in vivo with scanners used in clinical patient care. Clinically applied imaging techniques were utilised and optimised for small-animal imaging.
The selected human lung cancer cell lines are frequently used in preclinical xenograft research, typically on mice. In this study the authors decided to use a rat model. These animals can carry larger tumour sizes that mandate induction of relevant vascularisation to overcome restrictions of oxygen diffusion distance and additionally will develop relevant amounts of necrosis. Moreover, these tumour sizes are appropriate for evaluation with clinical scanners and imaging protocols. Two different lung cancer cell lines were utilised and both of them were investigated either with or without co-injection of vascular growth promoters (i.e. vascular growth factors and rat endothelial cells) creating two different levels of vascular development in the growing tumours of both cell lines. Since tumour vascularity in xenograft tumours is developed by the host animal, rat vascular growth factors were used to modify vascularity. Better tumour vascularisation causes a better supply of oxygen and nutrients which results in a faster growth rate. Nevertheless, growth rates are also dependent on the pathophysiological properties of the individual tumour cell line. Accordingly, after tumour and vascularity-dependent growth times tumours were evaluated and different levels of vascularity were quantifiable with clinical scanners and techniques.
Co-transplantation resulted in faster tumour growth. Since the amount of transplanted tumour cells was constant, the faster growth rate can be explained by a better supply of oxygen and nutrients as well as an improved evacuation of lactate and end products of metabolism due to better vascularisation. An important condition for fast tumour growth is the angiogenic switch that is dependent on endothelial precursor cells . Beside all other influences, co-transplantation with vascular growth factors and rat endothelial cells increased the probability of a successful switch to an angiogenic phenotype which led to better vascularisation in these tumours. This was proven in our study by the increases of MVD and the larger amount of embolised beads within the tumours. Interestingly, the co-transplanted tumours grew so fast that the development of central necrosis was higher in this group even though the number of larger vessels was also higher. In fact, CE-CT showed wide hypodense areas within a contrast-enhancing tumour parenchyma at the tumour rim and around larger central vessels in co-transplanted tumours. Using TOF sequences, co-transplanted tumours revealed the development of larger vessels in contrast to non-co-transplanted tumours. In summary, better vascularisation boosts fast growth which at the same time can increase necrosis by uncoupling regions from blood supply due to fast growth. Conversely, this means that regions of necrosis may be an indirect indicator of better vascularisation.
Adding a functional MRI technique, like quantitative DWI, with the calculation of ADC values, increases the diagnostic information. ADCdiff describes various properties of tumours. Higher ADCdiff values indicate higher diffusion; that is equating with fewer or less functional cell membranes frequently found in areas of necrosis or lower cellularity. As known from biology and neurology, necrosis is a continuum extending from viability to cell to tissue death. Therefore, a differentiated view on the ADCdiff is required since the latter allows for detecting different levels of necrosis.
Furthermore, ADCdiff conveys indirect information concerning the viable part of the tumours and the condition of the tumour microenvironment . The peak value of ADCdiff (high maximal ADCdiff) reflects the degree of necrosis, whereas the nadir of ADCdiff (low minimal ADCdiff) informs about the condition and cellular density of the viable tumour cells. Both permit direct and indirect interpretations on the level of vascularisation, blood supply, and the rate of growth. The range of the ADCdiff values provides an additional source of information for a differentiated interpretation of ADCdiff. The maximal ADCdiff in a tumour may reflect the extent of necrosis in certain regions. As mentioned before, better vascularisation induces faster tumour growth that may impede the balance between nutrient supply and nutrient demand which may trigger regional critical shortness in blood supply leading to increased necrosis [27, 28]. In addition, fast cellular growth rate is paralleled by the unorganised development of pathological vessels leading to critical shortage of oxygen in highly vascularised tumours. Such shortage leads to necrosis in relevant parts of larger tumours. The formation of regions with severe necrosis in better-vascularised tumours can be the consequence. Such necrosis is not seen in smaller tumours  because these tumours show a slower growth rate with more adequate vascular development and do not develop critically supplied areas . The histological findings in this study supported this fact and are in agreement with the current knowledge on the pathophysiology of tumour growth. In highly vascularised, large tumours, wide-diameter vessels, which lack further small branches, were found traversing through necrotic regions. Thus, nutrition in these regions was limited to diffusion from one large single vessel. In this respect, better-vascularised tumours exhibited areas with more necrosis. These areas were formerly viable tumour regions that lost their oxygen supply and became necrotic. This situation was found less often in slower-growing tumours. To sum up, tumours with better vascularisation exhibit both severe necrosis and well-vascularised areas. Consequently, the existence of large necrotic areas in a tumour can be an indirect indicator for good vascularisation. The differentiated view on ADCdiff could detect this indicator. More precisely, the maximal ADCdiff (high maximal values in a tumour) suggests the existence of areas of high necrosis which is more often found in better-vascularised tumours. Hence, a higher maximal ADCdiff suggests better-vascularised tumours. The results of this study support this fact. DWI is capable of detecting these differences and was already found to correlate with MVD . The ADCdiff is much higher in better-vascularised tumours because of the existence of highly necrotic areas exhibiting high diffusion values . The peak value of ADCdiff (maximal ADCdiff) indicates focal areas of severe necrosis found in tumours with faster growth due to better vascularisation. The latter was insufficient at some point of tumour growth resulting in critical shortage of oxygen and nutrients and consecutive development of necrosis. This may indicate a poorer prognosis as Marconi et al. recently showed in women with advanced-stage cervical cancer .
The interpretation of the minimal ADCdiff should also be meaningful. The two major factors for impeded diffusion as estimated by the lowest values of ADCdiff are the severely reduced intracellular space (due to more organelles and larger nuclei) and extracellular space (due to high proliferation rate) in viable tumour tissue. In general, the lower ADCdiff values correspond to higher cellularity and more compact tumour tissue. Impeded diffusion in tumours occurs if tumour cells can proliferate and form their typical tumour microenvironment. However, such proliferation requires oxygen and nutrient supply and, therefore, the existence of a vascular system within the tumour. As found herein, a low minimal ADCdiff may indirectly indicate well-vascularised areas. The available academic literature gathered contradictory results related to this topic. In women suffering from cervical cancer, Nakamura et al. found that low ADC values in the tumours were related to poor prognosis  which supports the findings in this study, since patients with highly vascularised tumours frequently exhibit a poorer prognosis.
In summary, the range of the ADCdiff is related to high vascularisation. More precisely, higher maximal ADCdiff values are associated with high necrosis as a consequence of very fast tumour growth while very low minimal ADCdiff values are associated with high cellularity, dense cell-cell contacts, and high protein expression as a consequence of good proliferation. Both fast growth and good proliferation may be surrogate parameters related to better vascularisation.
The ADCperf is capable of detecting perfusion in tissues directly and is related to vascularisation in tumours. In this tumour model, highly vascularised and well-perfused tumours exhibited higher ADCperf values, indicating increased tumour vascularisation. This was shown in both cell lines and was independent from the influence of the histological cell type. The results of this study show that the separation of the ADCs can provide physicians with more information. However, in addition to the suggestion by Koh et al.  that separating different b-values may contribute to this, it was shown herein that the differentiated analysis of minimal and maximal ADC values may increase the value of the imaging [34, 35].
DWI can be used in extension to histology and other imaging modalities. The evaluation of vascularisation in tumours can be improved if different methods are used in combination. ADCdiff, ADCperf, MVD, and other methods detect correlated characteristics of vascularisation in tumours. The low, but significant, correlation between MVD and maximal ADCperf supports the idea that both parameters are influenced by vascularisation properties. The reason for the comparatively weak correlation might be that both parameters are influenced by different aspects of vascularisation. MVD is predominantly influenced by the existence of vessels (which do not necessarily have to be well perfused), whereas the ADCperf is mainly influenced by the level of perfusion (which is not necessarily linked to the existence of intact vessel walls).
The correlation with the minimal ADCdiff values is also in line with this concept. Well-vascularised tumours developed areas with high proliferation and high cellularity which may account for the correlation of MVD and the minimal ADCdiff. This indicates that perfusion and vascularisation in tumours affect the DWI values obtained which was confirmed by the histological results of this study. Much more than the ADCperf, the minimal ADCdiff and the MVD are influenced by further aspects of vascularisation. The authors speculate that the minimal ADCdiff is linked indirectly to vascularisation, whereas MVD is connected much more closely to vascularisation. While this might explain the small magnitude of this correlation, a small, but significant, correlation can be also meaningful. As described before, highly vascularised fast-growing tumours reveal both well-perfused and necrotic areas. The correlation between CE-CT and DWI implies that both methods capture this situation. Necrotic areas were detected by high maximal ADCdiff values. CE-CT visualised larger, comparatively necrotic, volumes and stronger contrast enhancement in vascularised areas. In this case, both parameters – the maximal ADCdiff and the relative necrotic volume in CE-CT – are directly influenced by the same factor: the level of necrosis. Both diffusion and contrast enhancement increase with increasing levels of necrosis. Accordingly, a moderately significant correlation was found. Thus, DWI and CE-CT complement each other and extend the insights in the tumour biology concerning vascularisation.
One limitation of this study is the utilisation of clinical scanners for small-animal imaging with reduced resolution as compared to dedicated preclinical scanners. To overcome this disadvantage, a rat animal model was used that allows the use of higher tumour volumes. The frequently used mice tumour models only permit smaller tumour sizes for animal welfare reasons. Tumours of this size are difficult to scan with clinical scanners . In contrast, the rat tumours in this study were at least two centimetres in maximal extension. The pathophysiology related to vascularisation of tumours of this size may be much more comparable to tumours in patients since diffusion distances for oxygen and other nutrients cannot cover comparatively large tumour volumes, especially at the tumour margins. Tumour sizes investigated in this study are dependent on sufficient neovascularisation – as the tumours in patients are – and should provide information more comparable to the clinical situation. In diagnostic imaging of tumours of patients, the presence of large necrotic areas (hypodense in CE-CT in the venous phase) and highly vascularised peripheral areas (hyperdense in CE-CT in the arterial phase) is a frequent clinical finding. The applicability of clinical imaging in this model is demonstrated by the comparison to the histological and immunohistological results. Furthermore, even though we pushed the imaging protocols to their limits, we believe that the results found herein are a good basis for translating this to clinical patient care.
With respect to DWI applications in oncology, the reliable transferability of ADC values between different imaging systems has been under discussion for many years . The definition of standardised technical protocols for DWI and standardised interpretation of results are still under evaluation and increasing consensus is developing . In our study, we used standardised parameters that are applicable for clinical imaging and allow comparability to other imaging systems. Using the comparison with histology, this research added new information to the current research on how reliable ADC values for both diffusion and perfusion are. The use of two different cell lines each with two different stages of vascular development reduced the influences on imaging due to different histological traits of different cell lines like protein expression, size, and membrane permeability of the tumour cells. Consequently, differences in imaging parameters within the same tumour cell line are more likely caused by vascularisation rather than by characteristics of the cell line. This is an important advantage compared to other research investigating histologically diverse tumour lines. The investigated lung cancer cell lines are standard in preclinical research and are previously well described. Further clinical investigations will be developed taking advantage of the results found herein.
The present study established a human tumour xenograft cancer model with modifiable vascularisation and successfully evaluated parameters reflecting vascularisation in tumours by noninvasive clinical imaging. Results were correlated with histology and immunohistology. The amount of necrosis and the pathologic tumour vasculature in tumours were visualised using clinical scanners and were correlated with histological results. The results of the study showed that DWI is capable of assessing information not only on cell density but also indirectly on the vascularisation in tumours. The differentiated observation of DWI parameters increased the information obtainable from imaging and provide additional parameters that can be introduced to large-data analysis. ADCdiff should be interpreted in its minimal, mean and maximal extension. The harness of established clinical methods (DWI, CE-CT, TOF) may by expanded by a better understanding of these imaging parameters. The results of this study could contribute to this. This may give more information concerning the aggressiveness of tumours, which may influence therapy [4, 32, 33]. Additionally, this study showed that clinical imaging units may well be used for preclinical imaging, which allows faster translation of results.