Research ArticleExperimental Studies
Open Access

Correlation of Preclinical In Vivo Imaging Modalities and Immunohistochemistry for Tumor Hypoxia and Vasculature

REBECCA A. D’ALONZO, SYNAT KEAM, TRACY S. HOANG, SUKI GILL, PEJMAN ROWSHANFARZAD, ANNA K. NOWAK, ALISTAIR M. COOK and MARTIN A. EBERT
In Vivo January 2025, 39 (1) 55-79; DOI: https://doi.org/10.21873/invivo.13804

Abstract

Background/Aim: Tumors exhibit impaired blood flow and hypoxic areas, which can reduce the effectiveness of treatments. Characterizing these tumor features can inform treatment decisions, including the use of vasculature modulation therapies. Imaging provides insight into these characteristics, with techniques varying between clinical and preclinical settings. Materials and Methods: To investigate changes in different tumor regions over time, R2* values from blood oxygen-level dependent MRI (BOLD-MRI), blood flow from power Doppler ultrasound, and oxygen saturation from photoacoustic ultrasound were analyzed and compared to CD31+ and pimonidazole tissue staining. To aid in preclinical translation, the fluorescence of a hypoxia probe was also compared to ultrasound techniques. Results: The imaging techniques detected tumor heterogeneity and an overall decrease in blood flow and oxygen levels over time. The analysis found varying correlations between regions, indicating an indirect relationship between imaging outcomes, which is influenced by external factors. Regional analysis allowed for more accurate results, as areas less affected by various factors were examined separately from highly impacted regions, aiding in their identification. Conclusion: Examining tumor regions with multiple imaging techniques allowed for better understanding and identification of modality-specific limitations, as certain techniques may incorrectly suggest that tumors are more vascularized and less hypoxic than they are.

Key Words:

Abnormal tumor vasculature and the resulting hypoxia are critical aspects of the tumor microenvironment (TME). Vasculature is unevenly distributed throughout the tumor, with more blood vessels in the tumor periphery and fewer in the center (1, 2). This uneven distribution significantly influences tumor progression and treatment resistance (3, 4). Non-invasive imaging of these characteristics can be used to guide treatment decisions and aid in prognosis evaluation (5, 6). Identifying and quantifying these features allows monitoring of a patient’s response to treatments, particularly for therapies targeting tumor vasculature and hypoxia (7-9). Given the treatment resistance of some tumors, identifying patients who could benefit from TME modulation and assessing the effectiveness is crucial for improving outcomes. This would avoid the use of unnecessary and ineffective treatments, and therefore minimize the risk of adverse events and the severity of side effects. Although vasculature modulation techniques, such as anti-angiogenic drugs and radiotherapy, have shown promise in preclinical mesothelioma research, this success is yet to be translated into the clinic (3).

Imaging techniques can provide a comprehensive understanding of TME modulation, including spatial and temporal changes. Imaging plays a pivotal role in assessing the effectiveness of vasculature modulation, guiding future decisions. Preclinical testing of modulation therapies is essential, but there is currently a disconnect between clinical and preclinical imaging. It is crucial to understand the effect on both clinical and preclinical tumors, emphasizing the importance of translatable imaging techniques. In preclinical research, bioluminescence and fluorescence are commonly used, as they are cost-effective, fast, and can image multiple animals simultaneously (10). Numerous targeted probes have been developed that bind to various tumor factors. A common probe target is carbonic anhydrase IX (CAIX), a protein that is upregulated in hypoxic conditions (10, 11). However, no hypoxia specific probe has reached clinical use in solid tumors, though investigations are underway (11, 12). Other clinical probes are being developed that bind to different tumor targets, along with the investigation of autofluorescence (13, 14). Fluorescence techniques are limited by depth penetration and light absorption, so they have predominantly been applied to surface tumors, interoperative tumor margin identification, and some non-malignant conditions (11-16).

Ultrasound and optical techniques offer quick imaging times, therefore suitable for patients and preclinical models. There are several ultrasound methods that can be used to image the TME, including Doppler, photoacoustic, and contrast-enhanced ultrasound (CE-US) (10, 17). Ultrasound techniques are easily translated to the clinic (17, 18). Investigation into ultrasound techniques for preclinical lung cancer (19) and clinical pneumonia (17) have shown successful lung imaging despite respiratory and cardiac movements. Magnetic resonance imaging (MRI) techniques can be used to detect oxygen levels in both tissues (tissue oxygen-level dependent; TOLD) and blood (blood oxygen-level dependent; BOLD) (10, 20). Clinically, these techniques are mainly applied to image conditions of the head and neck (e.g., brain tumors and stroke) (21). This is partially due to several factors that these MRI techniques are susceptible to, such as movement and flow and oxygenation-dependent (FLOOD) effects (21, 22). Ongoing clinical studies aim to expand use to other locations with success improving. Clinical positron emission tomography (PET) imaging is common, using radiotracers that can bind to different TME components (10, 23). However, the limitations of PET, including instrument expenses, radiotracer production, and poor spatial resolution, have restricted its applicability in preclinical research (10). Both MRI and PET have prolonged imaging times, increasing the risk of adverse events in anesthetized animals.

Currently, no non-invasive in vivo imaging technique has been validated across both the clinical and preclinical settings that is readily and repeatedly usable, cost-effective, and has limited risks. Although tissue staining is considered the gold standard, it is not without limitations (10). This study aimed to investigate longitudinal changes in blood flow and hypoxia over time as detected by BOLD-MRI and ultrasound-based imaging techniques. Additionally, we investigated how well each modality can identify these characteristics spatially within the tumor. This is crucial for the detailed examination required for future work with TME modulators. We also aimed to correlate preclinical fluorescence to clinically relevant ultrasound techniques.

Materials and Methods

Mice. Female BALB/cJAusBP mice (aged 6-8 weeks) were obtained from Harry Perkins Institute of Medical Research (HPIMR) South Facility (Perth, WA, Australia) and housed at HPIMR North Facility (Perth, WA, Australia). All mouse experiments were approved by the respective authorities (AE178) and conducted in accordance with the Australian Code for the Care and Use of Animals for Scientific Purposes (8th edition, 2013) and University of Western Australia animal ethics guidelines and protocols.

Tumor model. AB1-HA cells were obtained from the National Centre for Asbestos Related Diseases (Perth, WA, Australia). AB1-HA cells were resuspended in sterile 1X phosphate-buffered saline (PBS) at 5×106 cells/ml, and 100 μl (5×105 cells) was subcutaneously injected into the right flank (designated ‘day 0’). Tumor width (W) and length (L) were measured using calipers, tumor volume (V) was calculated by V=(W2×L)/2.

Ultrasound imaging and analysis. Ultrasound imaging was performed using a Vevo LAZR-X High Frequency Ultrasound and Photoacoustic Imaging system (Fujifilm VisualSonics, Toronto, ON, Canada). The MX550D transducer was used for power Doppler (gain: 40 dB) and photoacoustic ultrasound imaging (gain: 40 dB). For CE-US, the MX250 transducer (gain: 30 dB) was used. The Vevo MicroMarker non-targeted oxygen-filled microbubble (Fujifilm VisualSonics, Toronto, ON, Canada) contrast agent was injected (40 μl) into the tail vein of mice at the time of imaging. A 3D scan of the whole tumor with a step size of 0.05 mm was taken. Mice were anesthetized with 1.5-2% isoflurane in 100 ml/min medical air for 20-30 min.

Ultrasound images were imported into VevoLab Analysis software (version 5.5.1) (Fujifilm VisualSonics). The maximum area of the tumor in the transverse plane was identified, and 5 slices on both sides were contoured. The software calculated the percentage of functional vasculature blood flow (%Vflow) from power Doppler images, the oxygen saturation in tissue (%sO2) and blood (%sO2Blood) from photoacoustic images, and the contrast agent percentage (%CA) in the tumor volume. The %Vflow and %sO2 difference between the weeks was calculated by subtracting the second week from the first week, giving the Δ%Vflow and Δ%sO2, respectively.

BOLD-MRI and analysis. BOLD-MRI was conducted on a 3T microMRI (MR Solutions, Guildford, UK). BOLD images were acquired using a multi-echo gradient echo sequence (MGEMS) with a repetition time (TR) of 150 ms, 10 echo delay times (TE) from 7 to 63 ms with echo spacing of 7 ms, flip angle (FA) of 20, and acquisition time 19 s. The sequence was run 15 times for each gas, with a single slice (thickness of 1.5 mm) in the transverse plane taken.

Mice acclimatized for 10 min after each gas change. Mice breathed 1.5-2% isoflurane mixed with 100 ml/min 100% oxygen (O2), then carbogen (CB; 95% O2 & 5% CO2). The gas was changed back to O2, and the sequence was run. Mice were anesthetized for 45 min. Values from the first CB and second O2 were used for analysis.

BOLD image analysis was conducted on ImageJ (version 1.54) (Fiji 2.15, National Institutes of Health, Bethesda, MD, USA). Regions were contoured, with measurements taken to ensure they matched ultrasound regions. Images were loaded into the MRI Analysis Calculator, the transverse relaxation rate corrected for external magnetic field inhomogeneity (T2*) values were output and the inverse corrected transverse relaxation rate (R2*) values calculated (R2*=1/T2*).

The change in R2* between CB and O2 inhalation (δR2*) was calculated using the formula:

Embedded Image

Where R2*CBavg is the mean R2* value during CB breathing, and R2*O2 are the individual values with O2 inhalation.

The difference in δR2* between the first and second week (ΔR2*) was calculated using the formula:

Embedded Image

Where δ1R2*avg is the mean difference between O2 and CB for the first week (calculated from Equation 1) and δ2R2*avg is the mean difference between O2 and CB for the second week (calculated from Equation 1).

Near-infrared fluorescence and analysis. Fluorescence imaging of the IVISense Hypoxia CAIX 680 probe (HypoxiSense 680; PerkinElmer Inc., Hopkinton, MA, USA) was conducted on a Caliper IVIS Lumina II in vivo imaging system (PerkinElmer Inc., Hopkinton, MA, USA). The HypoxiSense 680 probe was injected intraperitoneally (100 μl) into mice 24 h before imaging. Fluorescence images were taken at the following settings: 675 ex/ICG em for the probe and 500 ex/Cy55 em for the background (ex: excitation, em: emission), exposure time was auto, binning was medium, and FStop was 2. Mice were anesthetized with 1.5-2% isoflurane in 100 ml/min medical air for 15 min.

Both fluorescence and photographic images were taken and analyzed using the Living Image Software (version 4.2) (Caliper Life Sciences, Hopkinton, MA, USA). Images were loaded into spectral unmixing, and the targeted probe was selected for. The intensity of a fluorescence signal is presented as the total radiant efficiency (TRE) – the radiance of the subject per excitation intensity.

Immunohistochemistry. Mice were injected with pimonidazole (Hypoxyprobe, Hypoxyprobe Inc., Burlington, MA, USA) 60 min prior to euthanasia (intraperitoneal injection, 60 mg/kg) for immunohistochemistry (IHC) staining of the hypoxia marker. Tumors were collected and frozen in OCT compound buffer (Thermo Fisher Scientific, Waltham, MA, USA), positioned so sectioning aligned with the imaging plane. Sections 7 μm thick were stained with various antibodies (Table I) and DAPI nuclear stain.

View this table:
Table I.

Antibodies for immunohistochemistry staining.

Slides were examined by PANNORAMIC 250 Flash III slide scanner (3DHISTECH, Budapest, Hungary) and analyzed using QuantCenter (3DHISTECH). The length (μm) of conjoined CD31+ cells (vessels) was measured via manual drawing. For pimonidazole staining, the percentage of the area positive for the marker (%hypoxic area) was calculated using a mask on QuantCenter. At least three slices per tumor were analyzed.

Statistical analysis. A group of four mice provided 80% power (p=0.05) to detect a 40% change, considering a 20% standard deviation. Statistical analyses were performed using GraphPad Prism (version 9.5) (GraphPad Software, Inc., Boston, MA, USA). Statistical significance was set at p<0.05.

To detect differences in %Vflow and %sO2 between tumor regions in individual mice for the same imaging session, a one-way analysis of variance (ANOVA) with Tukey’s correction was conducted. To compare changes in %Vflow and %sO2 between the weeks, individual tumors underwent a one-way ANOVA with Šidák correction with region matching (the same region was compared across weeks). To detect significant differences in R2* values between O2 and CB inhalation, a one-way ANOVA with Šidák correction and region matching for each week was performed. To compare δR2* values (determined by Equation 1) between different regions at the same time point, a one-way ANOVA with Tukey’s correction was conducted. The δR2* between the weeks, were compared via one-way ANOVA with Šidák correction and region matching.

For CD31+ and pimonidazole IHC staining analysis, the vessel length in each region and the percentage of the hypoxic area (%hypoxic area) was compared using one-way ANOVA with Tukey’s correction, respectively.

The TRE, %Vflow, %sO2, %sO2Blood, and %CA for the first week were compared to the second via a two-tailed paired t-test with individual mice paired. Correlations were calculated using Pearson’s correlation coefficient.

Results

Analysis of ultrasound and BOLD-MRI show strong regional variations. The R2* change (δR2*) observed in BOLD images between inhalation of 100% O2 and CB gives an idea of tumor hypoxia (6, 20). A significant detectable difference in R2* between the two gases serves as a surrogate indicator that the tumor is not hypoxic, since absorbed blood gases are able to enter the tumor via circulation. Functional blood flow (%Vflow) was visualized by power Doppler ultrasound, and photoacoustic ultrasound detected the oxygen saturation in tumors (%sO2). To evaluate how well these techniques distinguish distinct tumor regions, mice were imaged according to the schedule in Figure 1A. Specific regions within the tumors were contoured (Figure 1B). Tumor growth curves are provided in Figure 1C.

Figure 1.
Figure 1.

Imaging schedule and region analysis contours. (A) Schematic showing the experimental timeline. Subcutaneous injection of AB1-HA mesothelioma cells occurred on day 0. Imaging commenced 12 days post-inoculation, when tumors were 75.8±26.4 mm3 (28.5±7.4 mm2; mean±standard deviation). The tumor volume on day 19 post-inoculation was 556.6±234.5 mm3 (105.6±29.0 mm2). On day 20 post-inoculation, all mice received an intraperitoneal injection of pimonidazole 60 min before euthanasia. The experiment was repeated once with four mice in each group (n=4 mice). (B) Representative images illustrating the regions for Week 1 for power Doppler with B-mode ultrasound and BOLD-MRI. The regions contoured are the Core (blue, tumor center), Region 1 (dark pink, superficial ventral), Region 2 (light purple, deep ventral), Region 3 (dark purple, deep dorsal), Region 4 (light pink, superficial dorsal), Region 5 (light green, tumor edge ventral), and Region 6 (dark green, tumor edge dorsal). The Core had a diameter of 1 mm for the first week and 2 mm for the second week. (C) Individual tumor growth curves for each mouse. The dotted lines indicate imaging sessions. The average change in size between imaging days in the same week were 15.53±7.24 mm3 (4.68±2.24 mm2) for the first week (days 12 & 13) and 129.67±86.01 mm3 (16.15±9.03 mm2) for the second week (days 19 & 20). BOLD-MRI: Blood oxygen-level dependent MRI.

During the first week, the tumor Core tended towards smaller %Vflow (Figure 2A) and %sO2 (Figure 2B) compared to other regions, however, exceptions did occur (%Vflow in Mouse 1; Figure 2A). Some regions had no significant difference in R2* between O2 and CB inhalation (Table II). Notably, Region 2 had the smallest δR2* for all mice, with a significant δR2* in Mouse 1 only (Table II). The small δR2* in Region 2 implies relatively lower oxygen flow compared to other regions. This is supported by ultrasound imaging, specifically, three tumors had no detectable %Vflow in Region 2 and the single tumor with %Vflow (Mouse 1; Figure 2A) was the same tumor with statistically significant δR2* (Table II). Furthermore, the %Vflow and %sO2 in Region 2 were significantly lower than those in most other regions (Table III). However, the relationship between %Vflow and %sO2 with δR2* was not always straightforward. For example, Region 3 exhibited less %Vflow and %sO2 than other regions, but it had the greatest δR2*, excluding Mouse 2 (Figure 2 and Table II). For the second week, there were less regions with significant δR2* compared to the first week (2-3 vs. 5-8 regions, respectively; Table II). Interestingly, compared to the first week, Region 2 now had a significant δR2* in two tumors (Table II), no/reduced %Vflow (Figure 2A), and %sO2 increased in half the tumors and decreased in the others (Figure 2B). This demonstrates that not all regions with low %Vflow and/or %sO2 had non-significant δR2*. For individual regions, there was a stronger correlation between %Vflow vs. δR2* than %sO2 vs. δR2* overall (Table IV).

Figure 2.
Figure 2.
Figure 2.

Change in functional vasculature, oxygen saturation, and δR2* in different tumor regions over time. (A) Analysis of power Doppler ultrasound determined the change in %Vflow in different contoured regions of the tumor volume between the first and second imaging weeks. Results show the mean %Vflow±95% confidence interval (95%CI) for each region at each time point. Mean %V flow values <1 are shown numerically. (B) Analysis of photoacoustic ultrasound determined the change in %sO2 in different contoured regions of the tumor volume between the first and second imaging weeks. Results show the mean %sO2±95%CI for each region at each time point. Mean %sO2 values <1 are shown numerically. (C) Analysis of BOLD-MRI determined the δR2* (the change in R2* between O2 and CB gas breathing) in different contoured regions of the tumor volume between the first and second imaging weeks. Results show the mean δR2*±SEM for each region for each time point. All p-values were calculated using one way ANOVA with Šidák correction and region matching. %Vflow: Functional vasculature; %sO2: oxygen saturation in tissue; BOLD-MRI: blood oxygen-level dependent MRI; δR2*: difference in R2* for gas inhalation; CB: carbogen.

View this table:
Table II.

BOLD-MRI R2* values for gas breathing.

View this table:
Table III.

Weekly intra-modality comparison between regions for power Doppler ultrasound, photoacoustic ultrasound, and BOLD-MRI.

Examining the changes between the first and second week, all mice had a significant decrease in %Vflow (Figure 2A) and %sO2 (Figure 2B) in the whole tumor, whereas the change in δR2* not significantly different (Figure 2C). Tumors generally exhibited greater δR2* for the first week (Figure 2C), indicating tumors were less hypoxic compared to the larger tumors of the second week. However, the number of regions with a significant difference in δR2* was highly varied (2-6 regions), and no particular region was consistently significant across all mice (Figure 2C). While most regions exhibited a decrease in %Vflow and %sO2, a few regions located at the edges of tumors experienced an increase in the second week, with Region 5 and Region 6 being the most common (Figure 2A and B). When comparing the changes across regions, Region 2 had the weakest correlation between Δ%sO2 vs. Δ%Vflow (r=−0.275, p=0.725; Figure 3A) and ΔR2* vs. Δ%Vflow (r=−0.507, p=0.493; Figure 3B). The widest variation in regional correlations was between Δ%sO2 vs. ΔR2*, having the greatest range (r=0.071 to 0.0895) with the weakest overall correlation (r=0.074, p=0.687; Figure 3C).

Figure 3.
Figure 3.

Regional correlations between the temporal changes in functional vasculature, oxygen saturation, and δR2*. Pearson correlations for the whole tumor (orange), Core (blue), Region 1 (dark pink), Region 2 (light purple), Region 3 (dark purple), Region 4 (light pink), Region 5 (light green), Region 6 (dark green), and all data combined (grey) for (A) Δ%sO2 vs. Δ%Vflow, (B) ΔR2* vs. Δ%Vflow, and (C) ΔR2* vs. Δ%sO2. Significance is influenced by the number of data points for each region (n=4), excluding the combined data (n=32). The lines are aids in visual interpretation of the correlation; they do not suggest a relationship. Δ%Vflow: Difference in functional vasculature over time; Δ%sO2; difference in oxygen saturation in tissue over time; ΔR2*: difference in δR2* over time.

View this table:
Table IV.

Regional correlations between power Doppler ultrasound, photoacoustic ultrasound, and BOLD-MRI have limited significance.

Comparison of immunohistochemistry to ultrasound and BOLD-MRI. Tissue staining is considered the gold standard for characterizing tumors. For this study, two IHC stains were used: CD31+ and pimonidazole. CD31+ staining revealed abnormal tumor blood vessels, characterized by long and thick appearance, while more functional blood vessels were shorter and narrower (Figure 4A) (24). Pimonidazole binds to hypoxic cells (10), with the percentage of area covered by the stain (%hypoxic area) indicating hypoxia (Figure 5A). For this section, Region 5 and Region 6 were excluded from analysis. Tumor edges were not as well-preserved during tissue processing. Exclusion ensures analysis is focused on well preserved tumor regions, enhancing the reliability and interpretability of the comparisons drawn between tissue staining and multimodal imaging data.

Figure 4.
Figure 4.

Vessel length (CD31+) for different tumor regions. (A) Representative IHC images of CD31+ (pink) showing the difference in the Core and peripheral regions (Region 2). DAPI staining is blue. (B) Results show the median vessel length±95%CI for different regions. The mean is indicated by the ‘+’. The p-values were calculated using one way ANOVA with Tukey’s correction. IHC: Immunohistochemistry.

Figure 5.
Figure 5.

Hypoxic area (pimonidazole) for different tumor regions. (A) Representative IHC images of pimonidazole staining (green) showing differences in the Core and peripheral regions (Region 2). DAPI staining is blue. (B) Results show the %hypoxic area±SEM for different regions. Each data point is the %hypoxic area for a tissue slice, the number of points (n=4-6) varies depending on the quality of that tissue section. The p-values were calculated using one way ANOVA with Tukey’s correction. IHC: Immunohistochemistry; %hypoxic area: percentage of pimonidazole positive area.

Analysis of endothelial cell staining (CD31+) revealed that the tumor Core exhibited significantly longer blood vessels compared to other regions (Figure 4B), indicating the blood vessels in the middle of the tumor are less functional compared to the tumor periphery. Contradictory to expectations, pimonidazole staining revealed the Core did not have a greater %hypoxic area (Figure 5B). All tumors had significantly less %hypoxic area in the Core compared to Region 1, however, no consistent relationship with other regions was observed (Figure 5B). Interestingly, the %hypoxic area in the Core was relatively similar across mice (12.08±1.35 %hypoxic area) compared to ranges seen in other regions. The relatively small hypoxic area in the center could be attributed to the inability of the pimonidazole to enter the tumor, supported by CD31+, ultrasound, and BOLD-MRI, which all indicate this area is hypoxic with limited blood flow. Regional comparisons support this interpretation. For example, the Core in Mouse 1 had significantly less %Vflow, %sO2, and δR2* than Region 1, yet the Core had significantly less %hypoxic area (Table III and Figure 5B). This pattern is seen in other tumors, such as Region 2 in Mouse 4. However, this association was not consistently observed across all regions and varied between imaging modalities. For instance, Region 1 in Mouse 2 showed significantly more %Vflow than the other peripheral regions, but it had significantly less %sO2, no significant difference in δR2*, and only significantly more %hypoxic area compared to Region 4 and the Core (Table III and Figure 5B).

Correlations with both the mean and median CD31+ vessel length were conducted, due to the range and distribution of lengths. Correlations with the mean and median vessel length were similar in all regions except Region 4, which was different across all comparisons (Figure 6A-D). There was a negative correlation between the mean vessel length vs. %Vflow for all regions except Region 1, and Region 1 and Region 4 for the median vessel length (Figure 6A). There was a positive correlation between %hypoxic area vs. %Vflow, excluding Region 2 and Region 4 (Figure 6E). These support the idea that functional vessels, indicated by higher %Vflow and smaller vessel length, are needed for pimonidazole to enter tumors. Most regions had a negative correlation between %hypoxic area vs. %sO2 (Figure 6F). However, the relationship of %sO2 with the mean and median vessel length had an even distribution of negative and positive correlations (Figure 6B). All regions had a positive correlation between δR2* vs. vessel length (Figure 6C), indicating smaller, more functional vessels allowed for better gas contrast and thus a larger (absolute) δR2*. However, there was no consistent relationship between δR2* vs. %hypoxic area (Figure 6G). There was a positive correlation between the %hypoxic area vs. CD31+ vessel length (excluding Region 4) with varying strengths (Figure 6D).

Figure 6.
Figure 6.
Figure 6.

Regional correlation between CD31+ staining, pimonidazole staining, and imaging modalities. Pearson correlations for the Core (orange), Region 1 (dark pink), Region 2 (light green), Region 3 (dark purple), Region 4 (light blue), and all data combined (grey) for the (A) vessel length vs. %Vflow, (B) vessel length vs. %sO2, (C) vessel length vs. δR2*, and (D) vessel length vs. %hypoxic area. Due to the range of vessel lengths, both the mean and median were analyzed. The solid circles and solid lines show the mean CD31+ vessel length, while the hollow circles and dotted lines show the median vessel length. (E) %hypoxic area vs. %Vflow, (F) %hypoxic area vs. %sO2, and (G) %hypoxic area vs. δR2*. Significance is influenced by the number of data points for each region (n=4), excluding the combined data (n=20). The lines are aids in visual interpretation of the correlation; they do not suggest a relationship. %Vflow: Functional vasculature; %sO2: oxygen saturation in tissue; δR2*: difference in R2* for gas inhalation; %hypoxic area: percentage of pimonidazole positive area.

Comparison of ultrasound and fluorescence showed good correlation. Various ultrasound techniques are used in both the clinical and preclinical settings (10, 17-19). Along with power Doppler (%Vflow) and photoacoustic ultrasound (%sO2 and %sO2Blood), CE-US measures perfusion at the capillary level via the percentage of the contrast agent (%CA) in the tumor. Fluorescence imaging of hypoxia used the CAIX targeted probe HypoxiSense 680 to determine the total radiant efficiency (TRE). By comparing and correlating the TRE probe to ultrasound techniques, we gain an understanding of how the modalities relate, therefore aiding in the translation of preclinical fluorescence to the clinic. To examine how well these techniques can detect changes in tumors over time, mice were imaged according to the schedule in Figure 1A.

There was a significant decrease in %Vflow (p=0.00473), %sO2 (p=0.00148), %sO2Blood (p=0.0475), and %CA (p=0.0032) in the second week compared to the first week across all mice (Figure 7A-D). All mice had a significant increase in the TRE (Figure 7E), suggesting the tumor had progressed to a more hypoxic state compared to the first week. To examine the relationship between TRE and ultrasound outcomes, correlation analysis was performed (Figure 8). None of the individual week correlations were significant. Negative correlations, as expected, were observed between the TRE vs. %Vflow for individual weeks (r=−0.662 and r=−0.907 for the first and second week, respectively; Figure 8A) and both weeks combined (r=−0.686, p=0.050; Figure 8A). The correlation between TRE vs. %sO2 exhibited larger variability, with opposing directions for individual weeks (r=0.196 and r=−0.109 for the first and second week, respectively; Figure 8B), however, combined data from both weeks resulted in a stronger correlation (r=−0.784, p=0.0212; Figure 8B). The overall correlation for TRE vs. %Vflow was closer to that of the individual weeks compared to that of TRE vs. %sO2. The %sO2Blood consistently demonstrated a strong and robust correlation with TRE for both weeks (r=−0.919 and r=−0.882 for the first and second week, respectively; Figure 8C), and when the weeks were combined (r=−0.917, p=0.0013; Figure 8C). The correlation between TRE vs. %CA for the second week was positive (r=0.9061; Figure 8D), which goes against expectations. The combined data from both weeks had a negative non-significant correlation for TRE vs. %CA (r=−0.6255, p=0.133; Figure 8D).

Figure 7.
Figure 7.

Change in functional vasculature, oxygen saturation, blood oxygen saturation, contrast agent, and TRE in different tumor regions over time. Analysis of (A) power Doppler ultrasound to determine the %Vflow, (B) photoacoustic ultrasound to determine the %sO2, (C) photoacoustic ultrasound to determine the %sO2Blood, (D) non-linear CE-US to determine the %CA, and (E) the TRE of HypoxiSense. Results show the mean±standard mean error (SEM), with data points for each mouse at each time point, n=4 mice (excluding %CA, n=3) The p-values were calculated using a two-tailed paired t-test. %Vflow: Functional vasculature; %sO2: oxygen saturation in tissue; %sO2Blood: oxygen saturation in blood; CE-US: contrast-enhanced ultrasound; %CA: percentage of contrast agent in tumor volume; TRE: total radiant efficiency.

Figure 8.
Figure 8.

Correlation between the total radiant efficiency and ultrasound measurements. Pearson correlations for the first week (light blue), second week (dark blue), and both weeks combined (green) for the (A) TRE vs. %Vflow, (B) TRE vs. %sO2, (C) TRE vs. %sO2Blood, and (D) TRE vs. %contrast agent. The lines are aids in visual interpretation of the correlation; they do not suggest a relationship. TRE: Total radiant efficiency; %Vflow: functional vasculature; %sO2: oxygen saturation in tissue; %sO2Blood: oxygen saturation in blood; %CA: percentage of contrast agent in tumor volume.

Discussion

Imaging tumor vasculature and hypoxia can assist in treatment decisions, especially in the use and monitoring of TME modulatory therapies (7-9). Imaging techniques need to have spatial resolution that is sensitive enough to distinguish between different areas of tumors, as it is critical that the middle of tumors are altered for effective treatment. However, repeated imaging is also needed, limiting the use of radiation-based imaging modalities such as PET. Patient welfare and comfort also need to be considered, as long imaging times of MRI techniques could impact individuals. While commonly used in the clinic, PET and MRI have limitations when applied to preclinical research (10). Therefore, clinically relevant imaging modalities need to be compared to preclinical imaging.

We compared the temporal and spatial resolution of BOLD-MRI and ultrasound techniques. Analysis revealed the relationship between these modalities is region dependent, as each technique is influenced differently by factors. Most regions had a decrease in %Vflow and %sO2 over time; however, some peripheral regions had an increase (Figure 2A and B). A study by Drzał et al. used diffusion-weighted MRI (DW-MRI) to assess the perfusion in tumors at 8-, 13-, and 20-days post-inoculation (1). They found an overall significant decrease in perfusion in tumors over time, however, spatial analysis found significant heterogeneity of tumor perfusion. Hotspots of increased perfusion around the tumor periphery were observed as the tumor grew (1, 25). This observation reflects our results, where there was an increase in %Vflow and %sO2 at the edges of tumors (Figure 2A and B). Tumors that had an increase in %Vflow (Mouse 2 Region 6 and Mouse 3 Region 3) did not have an increase in %sO2, and those with an increase in %sO2 in several regions (Mouse 4) did not correspond to a change in %Vflow (Figure 2A and B). Together, these observations suggest that the increase in %sO2 is not solely due to existing blood vessels or new blood vessels forming in the outer areas as the tumor grows, but also due to oxygen diffusing throughout (26). It is still important to note that slower growing tumors are more likely to have vasculature growth that can match the tumor growth rate, which can still be abnormal with limited function (27).

Absolute R2* values have complex relationships with tumor oxygenation as they are influenced by multiple factors, thus the gas challenge is used as a contrast (6, 10, 20, 28, 29). The change in R2* due to the gases (δR2*) has been found to strongly correlate with the partial pressure of oxygen (pO2) but a universal quantitative relationship is yet to be derived (6, 10, 20, 28). However, the relationship remains that the more oxygenated a tumor is, the smaller the absolute R2* and the greater the δR2* (9, 28, 29). Multiple studies have found a strong correlation between BOLD-MRI R2* response and TOLD-MRI R1 response, and R1 has a higher correlation with pO2 (6, 28, 29).

Contrary to expectations, we found that the Core δR2* was significant in all mice for the first week, while the δR2* in some peripheral regions were not (Table II). We also found smaller absolute R2* values for the Core compared to those for most peripheral regions for the first week (Table II). This was also seen by Zhou et al., where the central tumor had smaller R2* values than the periphery (29). These somewhat unexpected results of a lower R2* in some hypoxic regions are likely influenced by factors that BOLD-MRI is sensitive to such as air flow rate (22, 28, 30). We found that ΔR2* for individual regions had a better correlation with Δ%Vflow than Δ%sO2 (Figure 3), suggesting blood flow has a closer relationship to BOLD-MRI than oxyhemoglobin/deoxyhemoglobin. Some studies have found that BOLD-MRI, which reflects vasculature oxygenation, can occasionally have a contradictory relationship with tumor oxygenation depending on conditions (28, 31).

The relationship between the different imaging outcomes is unclear and can vary highly depending on the region of the tumor and the tumor itself (27). Discrete areas of the tumor are subjected to external factors that impact them differently, thus adding to the complexity, as seen by the poor global correlations and large variation between individual tumors. Some imaging methods are more sensitive to external factors, such as anesthetic and air flow. A study by Joseph et al. found that spleen %sO2 could vary by 30% depending on isoflurane concentration, as detected by photoacoustic imaging on a multispectral optoacoustic tomography device (32). They also found changes in the %sO2 and hemoglobin concentration over time, with organ dependent rates (32). Similar observations have been made in rat brains using functional MRI (33) and with injectable anesthetic (27, 34). This has implications for modalities with extended imaging times. The longer the animal is under, the slower the breathing rate, and therefore the anesthetic concentration must change. It has consequences for the long imaging times of functional MRI techniques, which could interfere with the results, although fluctuations are not always statistically significant (30). Changes in breathing are harder to detect during small animal MRI, as all monitoring is conducted via sensing equipment (e.g., heart rate and breathing monitors) as visual examination is not possible. This is further affected by difficulty in detecting movement during processing and analysis, whereas motion in ultrasound is easily seen. Given the real-time nature and short duration of image acquisition with ultrasound, it is possible to ask human patients to hold their breath to further reduce artefacts (17). It is also important to note that the ultrasound data was acquired from 3D images, and the slices analyzed therefore represent a larger portion of the whole tumor, whereas BOLD-MRI was limited to a single slice through a central region.

We found a lack of pimonidazole in the central region of the tumor, possibly due to the inability to access hypoxic regions due to dysfunctional blood vessels. This issue has been seen in other studies that use an exogenous contrast (29, 35). Zhou et al. found a similar issue with the use of the PET radiotracer 18F-FMISO, a nitroimidazole with similar biochemical properties to pimonidazole (29). The high %hypoxic area of the periphery seemingly contradicts the increase in %sO2 seen in some regions. There was also a positive correlation between %Vflow vs. %hypoxic area (Figure 6). While few correlations were significant, this helps support the notion that the limited blood vessels inhibit pimonidazole infiltration into the tumor, thus it becomes limited to and accumulates in the periphery. An example of this can be seen in Mouse 4, where Region 1 and Region 4 are the only ones with any %Vflow yet have the highest %hypoxic area. We found that δR2* vs. vessel length (CD31+) had a positive correlation for all regions (Figure 6). While none of the individual regional correlations were significant themselves, these were the only pair of variables that correlated in a directionally consistent manner for all regions. This overall trend may suggest a stronger relationship between CD31+ IHC and BOLD-MRI compared to other imaging techniques. Other studies have shown that R2* has a good correlation with CD31+ (36).

Tumor dimensions and vessel morphology play a role in the distribution of pimonidazole. In general, the largest tumor should have the greatest %hypoxic area, though variations in shape will lead to departures from this rule. For example, Mouse 2 had the largest volume (Figure 1C), yet less %hypoxic area than the smallest tumor (Mouse 4) for Region 4 and Region 1. The tumor from Mouse 2 was elongated in the coronal plane, resulting in a shallower tumor with a larger surface area exposed to the skin. Therefore, allowing for more oxygen diffusion to these superficial regions. The limitations of pimonidazole have been seen in other studies, with high intra-tumor and inter-tumor variability being reported (20, 26). Despite the limitations, pimonidazole has shown strong correlations with photoacoustic techniques (37, 38), and to a lesser extent with BOLD-MRI (39), with tumor shape being an important factor (40).

Tissue staining can have varied results, depending upon the place in the tumor the slice was obtained. Careful consideration was taken to ensure the middle of the tumor was correctly identified and sectioned. The dependence of staining on a particular location within the tumor reflects one of the limitations of tissue samples and biopsies. While pathological staining is routinely used, it only allows for a small region of tissue to be tested. It is also important to remember that the markers used in tissue staining are commonly used to identify structural changes, such as CD31+, and downstream factors, such as HIF-1α. While these are useful, they may not reflect the timing of actual functional changes. It is the functional changes that may have the greatest predictive potential.

A pixel-by-pixel comparison of ultrasound images, BOLD-MRI, and IHC was not possible. Direct registration and resampling were not possible given the nature of the imaging files created by the different instruments and software. There were also some restrictions applied to the images by proprietary software. As a result, registration of the contours was dependent on the individual’s accuracy. Despite meticulous measurements and careful contour matching, there are inherent errors in this method. This impacts the statistical power, especially for the larger contours of the second week (35). There were cost-benefit considerations when choosing region layouts; larger contour areas have more accurate co-registration but at the cost of spatial resolution. Larger contoured regions also reduce the impact that pixelwise fluctuations may experience. Repeating measurements on the same imaging set with different investigators and imaging additional tumors would help improve some of these issues. The high number of regions with no detectable %Vflow impacts the accuracy of the correlation and the extent of comparisons that can be made. To better translate the imaging techniques to the clinic, more tumor cell types at more time points would need to be analyzed to overcome the limitation of the ‘zeroes’, as different tumor types have been shown distinctively different anatomical and functional properties throughout the tumor (27, 41).

We found that in vivo near-infrared fluorescence and some ultrasound modalities generally had a good correlation (Figure 8). However, there were differences between the first and second weeks. Both the %sO2 and %CA had an unexpected positive correlation with the TRE for the second week (Figure 8B and C). The morphology of the tumor impacts its blood vessels, and therefore some tumors may have more capillaries in the periphery as it grows, leading to more contrast agent in those areas. This would influence the %CA detected in the tumor overall. The fluorescence probe binds to CAIX, which is a marker of hypoxic cells, but there was a more consistent correlation with blood vessels (%Vflow) than %sO2 (r=−0.662 to −0.907 vs. r=0.109 to −0.784, respectively; Figure 8A and B). This suggests that the ability of the probe to distribute throughout tumors is affected by the underlying vasculature. This trend is seen throughout this study, where the ability of markers, such as pimonidazole, to infiltrate and bind to hypoxic cells is restricted by blood flow into the associated tumor region. While in vivo fluorescence does not have a clinical equivalent for imaging solid tumors, there are positive attributes that make it ideal for preclinical research (10). Mesothelioma, along with numerous other cancer types, have high expression of CAIX, therefore the preclinical CAIX probe, HypoxiSense 680, is appropriate (42). However, to develop a reliable translation of fluorescence measurements, a larger sample size, different cell lines, and more time points would need to be included in the imaging studies. Any translation to patients via another imaging technique needs to be investigated further.

Conclusion

Our observations emphasize the heterogeneous nature of the tumor macroenvironment and microenvironment. We detected an overall decrease in blood flow and oxygen levels in tumors over time, however, specific outer regions exhibited an increase in these features. We found a complex relationship between the measurements from each imaging technique, with varying correlation between regions, and various time course changes in comparison to the whole tumor. These correlations also differed across the imaging sessions as the tumor grew. It is crucial to acknowledge that multiple factors can exert distinct and different effects on each imaging technique, such as anesthetic flow rates on BOLD-MRI, and tumor morphology on pimonidazole perfusion. Although identifying the exact impact of these factors is difficult, they should be considered during image analysis and comparisons. The significance of conducting regional analysis becomes clear, allowing for more accurate results, as less affected areas are examined separately from highly impacted regions, aiding in their identification. These findings demonstrate the importance of detecting and distinguishing the heterogeneity within tumors. Recognizing and understanding these differences is important, as certain techniques may falsely indicate a tumor is more vascularized and less hypoxic than it is. Therefore, examination of different tumor regions proves more valuable when informing treatment decisions.

Acknowledgements

The Authors acknowledge the contribution of Kirsty Richardson. The Authors acknowledge the facilities and scientific and technical assistance offered, at the Centre for Microscopy, Characterization and Analysis, University of Western Australia, with special thanks to Diana Patalwala and Ivan Lozić. Rebecca D’Alonzo was supported by the Sir Charles Gairdner Hospital Radiation Oncology Scholarship in Radiobiology. The National Centre for Asbestos Related Diseases was supported by the National Health and Medical Research Council with a Centre of Research Excellence grant, application ID: APP1197652. This study was supported by grant 1163065 from the Cancer Australia Priority-driven Collaborative Cancer Research Scheme. The contents are solely the responsibility of the individual Authors and do not reflect the views of Cancer Australia.

Footnotes

  • Authors’ Contributions

    Conceptualization: MAE, AMC, AKN, RAD. Methodology: MAE, AMC, RAD. Investigation: RAD, SK, TSH. Data curation: RAD, SK, TSH. Formal analysis: RAD. Visualization: RAD. Funding acquisition: MAE, AMC, AKN, SG. Project administration: MAE, AMC. Supervision: MAE, AMC, PR SG, AKN. Writing - original draft: RAD. Writing - review and editing: RAD, SK, TSH, MAE, AMC, SG, PR, AKN.

  • Funding

    This work was supported by the National Health & Medical Research Council with a Centre of Research Excellence grant APP1197652 (ANK, AMC), the Cancer Australia Priority-driven Collaborative Cancer Research Scheme grant 1163065 (MAE, AMC, AKN, SG), the Mesothelioma Applied Research Foundation grant (AMC), and the iCare Dust Diseases Board Fellowship (AMC).

  • Conflicts of Interest

    The Authors declare no potential conflicts of interest in relation to this study.

  • Received October 10, 2024.
  • Revision received October 21, 2024.
  • Accepted October 22, 2024.

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).

References

  1. ↵
    1. Drzał A,
    2. Jasiński K,
    3. Gonet M,
    4. Kowolik E,
    5. Bartel Ż,
    6. Elas M
    : MRI and US imaging reveal evolution of spatial heterogeneity of murine tumor vasculature. Magn Reson Imaging 92: 33-44, 2022. DOI: 10.1016/j.mri.2022.06.003
    OpenUrlCrossRefPubMed
  2. ↵
    1. Forster JC,
    2. Harriss-Phillips WM,
    3. Douglass MJ,
    4. Bezak E
    : A review of the development of tumor vasculature and its effects on the tumor microenvironment. Hypoxia (Auckl) 5: 21-32, 2017. DOI: 10.2147/HP.S133231
    OpenUrlCrossRefPubMed
  3. ↵
    1. Nowak AK,
    2. Brosseau S,
    3. Cook A,
    4. Zalcman G
    : Antiangiogeneic strategies in mesothelioma. Front Oncol 10: 126, 2020. DOI: 10.3389/fonc.2020.00126
    OpenUrlCrossRefPubMed
  4. ↵
    1. Martin JD,
    2. Seano G,
    3. Jain RK
    : Normalizing function of tumor vessels: progress, opportunities, and challenges. Annu Rev Physiol 81: 505-534, 2019. DOI: 10.1146/annurev-physiol-020518-114700
    OpenUrlCrossRefPubMed
  5. ↵
    1. White DA,
    2. Zhang Z,
    3. Li L,
    4. Gerberich J,
    5. Stojadinovic S,
    6. Peschke P,
    7. Mason RP
    : Developing oxygen-enhanced magnetic resonance imaging as a prognostic biomarker of radiation response. Cancer Lett 380(1): 69-77, 2016. DOI: 10.1016/j.canlet.2016.06.003
    OpenUrlCrossRefPubMed
  6. ↵
    1. Hallac RR,
    2. Zhou H,
    3. Pidikiti R,
    4. Song K,
    5. Stojadinovic S,
    6. Zhao D,
    7. Solberg T,
    8. Peschke P,
    9. Mason RP
    : Correlations of noninvasive BOLD and TOLD MRI with pO2 and relevance to tumor radiation response. Magn Reson Med 71(5): 1863-1873, 2014. DOI: 10.1002/mrm.24846
    OpenUrlCrossRefPubMed
  7. ↵
    1. Orlova A,
    2. Pavlova K,
    3. Kurnikov A,
    4. Maslennikova A,
    5. Myagcheva M,
    6. Zakharov E,
    7. Skamnitskiy D,
    8. Perekatova V,
    9. Khilov A,
    10. Kovalchuk A,
    11. Moiseev A,
    12. Turchin I,
    13. Razansky D,
    14. Subochev P
    : Noninvasive optoacoustic microangiography reveals dose and size dependency of radiation-induced deep tumor vasculature remodeling. Neoplasia 26: 100778, 2022. DOI: 10.1016/j.neo.2022.100778
    OpenUrlCrossRefPubMed
    1. Haedicke K,
    2. Agemy L,
    3. Omar M,
    4. Berezhnoi A,
    5. Roberts S,
    6. Longo-Machado C,
    7. Skubal M,
    8. Nagar K,
    9. Hsu HT,
    10. Kim K,
    11. Reiner T,
    12. Coleman J,
    13. Ntziachristos V,
    14. Scherz A,
    15. Grimm J
    : High-resolution optoacoustic imaging of tissue responses to vascular-targeted therapies. Nat Biomed Eng 4(3): 286-297, 2020. DOI: 10.1038/s41551-020-0527-8
    OpenUrlCrossRefPubMed
  9. ↵
    1. Liang J,
    2. Cheng Q,
    3. Huang J,
    4. Ma M,
    5. Zhang D,
    6. Lei X,
    7. Xiao Z,
    8. Zhang D,
    9. Shi C,
    10. Luo L
    : Monitoring tumour microenvironment changes during anti-angiogenesis therapy using functional MRI. Angiogenesis 22(3): 457-470, 2019. DOI: 10.1007/s10456-019-09670-4
    OpenUrlCrossRefPubMed
  10. ↵
    1. D’Alonzo RA,
    2. Gill S,
    3. Rowshanfarzad P,
    4. Keam S,
    5. MacKinnon KM,
    6. Cook AM,
    7. Ebert MA
    : In vivo noninvasive preclinical tumor hypoxia imaging methods: a review. Int J Radiat Biol 97(5): 593-631, 2021. DOI: 10.1080/09553002.2021.1900943
    OpenUrlCrossRefPubMed
  11. ↵
    1. Chen KT,
    2. Seimbille Y
    : New developments in carbonic anhydrase IX-targeted fluorescence and nuclear imaging agents. Int J Mol Sci 23(11): 6125, 2022. DOI: 10.3390/ijms23116125
    OpenUrlCrossRefPubMed
  12. ↵
    1. Janoniene A,
    2. Petrikaite V
    : In search of advanced tumor diagnostics and treatment: achievements and perspectives of carbonic anhydrase IX targeted delivery. Mol Pharm 17(6): 1800-1815, 2020. DOI: 10.1021/acs.molpharmaceut.0c00180
    OpenUrlCrossRef
  13. ↵
    1. Sieroń A,
    2. Sieroń-Stołtny K,
    3. Kawczyk-Krupka A,
    4. Latos W,
    5. Kwiatek S,
    6. Straszak D,
    7. Bugaj AM
    : The role of fluorescence diagnosis in clinical practice. Onco Targets Ther 6: 977-982, 2013. DOI: 10.2147/OTT.S42074
    OpenUrlCrossRefPubMed
  14. ↵
    1. Jeremiasse B,
    2. van den Bosch CH,
    3. Wijnen MWHA,
    4. Terwisscha van Scheltinga CEJ,
    5. Fiocco MF,
    6. van der Steeg AFW
    : Systematic review and meta-analysis concerning near-infrared imaging with fluorescent agents to identify the sentinel lymph node in oncology patients. Eur J Surg Oncol 46(11): 2011-2022, 2020. DOI: 10.1016/j.ejso.2020.07.012
    OpenUrlCrossRefPubMed
    1. Wang J,
    2. Fang R,
    3. Wang L,
    4. Chen G,
    5. Wang H,
    6. Wang Z,
    7. Zhao D,
    8. Pavlov V,
    9. Kabirov I,
    10. Wang Z,
    11. Guo P,
    12. Peng L,
    13. Xu W
    : Identification of carbonic anhydrase IX as a novel target for endoscopic molecular imaging of human bladder cancer. Cell Physiol Biochem 47(4): 1565-1577, 2018. DOI: 10.1159/000490875
    OpenUrlCrossRefPubMed
  16. ↵
    1. Hirano F,
    2. Yokoyama-Kokuryo W,
    3. Yamazaki H,
    4. Tsutsumino M,
    5. Sakai R,
    6. Satoh S,
    7. Kimura T,
    8. Tojo N,
    9. Kohsaka H,
    10. Harigai M
    : Comparison of fluorescence optical imaging, ultrasonography and clinical examination with magnetic resonance imaging as a reference in active rheumatoid arthritis patients. Immunol Med 41(2): 75-81, 2018. DOI: 10.1080/13497413.2018.1481578
    OpenUrlCrossRefPubMed
  17. ↵
    1. Abdel Kader M,
    2. Osman NMM
    : Implementation of chest ultrasound with color Doppler in diagnosis of pneumonia in adults. Egypt J Radiol Nucl Med 47(3): 771-781, 2016. DOI: 10.1016/j.ejrnm.2016.05.015
    OpenUrlCrossRef
  18. ↵
    1. Heijblom M,
    2. Piras D,
    3. Brinkhuis M,
    4. van Hespen JC,
    5. van den Engh FM,
    6. van der Schaaf M,
    7. Klaase JM,
    8. van Leeuwen TG,
    9. Steenbergen W,
    10. Manohar S
    : Photoacoustic image patterns of breast carcinoma and comparisons with Magnetic Resonance Imaging and vascular stained histopathology. Sci Rep 5: 11778, 2015. DOI: 10.1038/srep11778
    OpenUrlCrossRefPubMed
  19. ↵
    1. Raes F,
    2. Sobilo J,
    3. Le Mée M,
    4. Rétif S,
    5. Natkunarajah S,
    6. Lerondel S,
    7. Le Pape A
    : High resolution ultrasound and photoacoustic imaging of orthotopic lung cancer in mice: new perspectives for onco-pharmacology. PLoS One 11(4): e0153532, 2016. DOI: 10.1371/journal.pone.0153532
    OpenUrlCrossRefPubMed
  20. ↵
    1. O’Connor JPB,
    2. Robinson SP,
    3. Waterton JC
    : Imaging tumour hypoxia with oxygen-enhanced MRI and BOLD MRI. Br J Radiol 92(1095): 20180642, 2019. DOI: 10.1259/bjr.20180642
    OpenUrlCrossRefPubMed
  21. ↵
    1. Gauthier CJ,
    2. Fan AP
    : BOLD signal physiology: Models and applications. NeuroImage 187: 116-127, 2019. DOI: 10.1016/j.neuroimage.2018.03.018
    OpenUrlCrossRefPubMed
  22. ↵
    1. Howe FA,
    2. Robinson SP,
    3. McIntyre DJ,
    4. Stubbs M,
    5. Griffiths JR
    : Issues in flow and oxygenation dependent contrast (FLOOD) imaging of tumours. NMR Biomed 14(7-8): 497-506, 2001. DOI: 10.1002/nbm.716
    OpenUrlCrossRefPubMed
  23. ↵
    1. Nguyen AT,
    2. Kim HK
    : Recent developments in PET and SPECT radiotracers as radiopharmaceuticals for hypoxia tumors. Pharmaceutics 15(7): 1840, 2023. DOI: 10.3390/pharmaceutics15071840
    OpenUrlCrossRefPubMed
  24. ↵
    1. He B,
    2. Johansson-Percival A,
    3. Backhouse J,
    4. Li J,
    5. Lee GYF,
    6. Hamzah J,
    7. Ganss R
    : Remodeling of metastatic vasculature reduces lung colonization and sensitizes overt metastases to immunotherapy. Cell Rep 30(3): 714-724.e5, 2020. DOI: 10.1016/j.celrep.2019.12.013
    OpenUrlCrossRefPubMed
  25. ↵
    1. Chen JJ,
    2. Fu SY,
    3. Chiang CS,
    4. Hong JH,
    5. Yeh CK
    : Characterization of tumor vasculature distributions in central and peripheral regions based on Doppler ultrasound. Med Phys 39(12): 7490-7498, 2012. DOI: 10.1118/1.4762683
    OpenUrlCrossRefPubMed
  26. ↵
    1. Lowerison MR,
    2. Huang C,
    3. Lucien F,
    4. Chen S,
    5. Song P
    : Ultrasound localization microscopy of renal tumor xenografts in chicken embryo is correlated to hypoxia. Sci Rep 10(1): 2478, 2020. DOI: 10.1038/s41598-020-59338-z
    OpenUrlCrossRefPubMed
  27. ↵
    1. Martinho Costa M,
    2. Shah A,
    3. Rivens I,
    4. Box C,
    5. O’Shea T,
    6. Papaevangelou E,
    7. Bamber J,
    8. Ter Haar G
    : Quantitative photoacoustic imaging study of tumours in vivo: Baseline variations in quantitative measurements. Photoacoustics 13: 53-65, 2018. DOI: 10.1016/j.pacs.2018.12.002
    OpenUrlCrossRefPubMed
  28. ↵
    1. Zhao D,
    2. Pacheco-Torres J,
    3. Hallac RR,
    4. White D,
    5. Peschke P,
    6. Cerdán S,
    7. Mason RP
    : Dynamic oxygen challenge evaluated by NMR T1 and T2*—insights into tumor oxygenation. NMR Biomed 28(8): 937-947, 2015. DOI: 10.1002/nbm.3325
    OpenUrlCrossRefPubMed
  29. ↵
    1. Zhou H,
    2. Chiguru S,
    3. Hallac RR,
    4. Yang D,
    5. Hao G,
    6. Peschke P,
    7. Mason RP
    : Examining correlations of oxygen sensitive MRI (BOLD/TOLD) with [18F]FMISO PET in rat prostate tumors. Am J Nucl Med Mol Imaging 9(2): 156-167, 2019.
    OpenUrlPubMed
  30. ↵
    1. Gonçalves MR,
    2. Johnson SP,
    3. Ramasawmy R,
    4. Pedley RB,
    5. Lythgoe MF,
    6. Walker-Samuel S
    : Decomposition of spontaneous fluctuations in tumour oxygenation using BOLD MRI and independent component analysis. Br J Cancer 113(8): 1168-1177, 2015. DOI: 10.1038/bjc.2015.270
    OpenUrlCrossRefPubMed
  31. ↵
    1. Jordan BF,
    2. Crokart N,
    3. Baudelet C,
    4. Cron GO,
    5. Ansiaux R,
    6. Gallez B
    : Complex relationship between changes in oxygenation status and changes in R: The case of insulin and NS-398, two inhibitors of oxygen consumption. Magn Reson Med 56(3): 637-643, 2006. DOI: 10.1002/mrm.20963
    OpenUrlCrossRefPubMed
  32. ↵
    1. Joseph J,
    2. Tomaszewski MR,
    3. Quiros-Gonzalez I,
    4. Weber J,
    5. Brunker J,
    6. Bohndiek SE
    : Evaluation of precision in optoacoustic tomography for preclinical imaging in living subjects. J Nucl Med 58(5): 807-814, 2017. DOI: 10.2967/jnumed.116.182311
    OpenUrlAbstract/FREE Full Text
  33. ↵
    1. Masamoto K,
    2. Fukuda M,
    3. Vazquez A,
    4. Kim SG
    : Dose-dependent effect of isoflurane on neurovascular coupling in rat cerebral cortex. Eur J Neurosci 30(2): 242-250, 2009. DOI: 10.1111/j.1460-9568.2009.06812.x
    OpenUrlCrossRefPubMed
  34. ↵
    1. Baudelet C,
    2. Gallez B
    : Effect of anesthesia on the signal intensity in tumors using BOLD-MRI: Comparison with flow measurements by Laser Doppler flowmetry and oxygen measurements by luminescence-based probes. Magn Reson Imaging 22(7): 905-912, 2004. DOI: 10.1016/j.mri.2004.02.005
    OpenUrlCrossRefPubMed
  35. ↵
    1. Shah A,
    2. Bush N,
    3. Box G,
    4. Eccles S,
    5. Bamber J
    : Value of combining dynamic contrast enhanced ultrasound and optoacoustic tomography for hypoxia imaging. Photoacoustics 8: 15-27, 2017. DOI: 10.1016/j.pacs.2017.08.001
    OpenUrlCrossRefPubMed
  36. ↵
    1. Liang J,
    2. Ma R,
    3. Chen H,
    4. Zhang D,
    5. Ye W,
    6. Shi C,
    7. Luo L
    : Detection of hyperacute reactions of desacetylvinblastine monohydrazide in a xenograft model using intravoxel incoherent motion DWI and R2* mapping. Am J Roentgenol 212(4): 717-726, 2019. DOI: 10.2214/AJR.18.20517
    OpenUrlCrossRefPubMed
  37. ↵
    1. Quiros-Gonzalez I,
    2. Tomaszewski MR,
    3. Golinska MA,
    4. Brown E,
    5. Ansel-Bollepalli L,
    6. Hacker L,
    7. Couturier DL,
    8. Sainz RM,
    9. Bohndiek SE
    : Photoacoustic tomography detects response and resistance to bevacizumab in breast cancer mouse models. Cancer Res 82(8): 1658-1668, 2022. DOI: 10.1158/0008-5472.CAN-21-0626
    OpenUrlCrossRefPubMed
  38. ↵
    1. Gerling M,
    2. Zhao Y,
    3. Nania S,
    4. Norberg KJ,
    5. Verbeke CS,
    6. Englert B,
    7. Kuiper RV,
    8. Bergström A,
    9. Hassan M,
    10. Neesse A,
    11. Löhr JM,
    12. Heuchel RL
    : Real-time assessment of tissue hypoxia in vivo with combined photoacoustics and high-frequency ultrasound. Theranostics 4(6): 604-613, 2014. DOI: 10.7150/thno.7996
    OpenUrlCrossRefPubMed
  39. ↵
    1. O’Connor JP,
    2. Boult JK,
    3. Jamin Y,
    4. Babur M,
    5. Finegan KG,
    6. Williams KJ,
    7. Little RA,
    8. Jackson A,
    9. Parker GJ,
    10. Reynolds AR,
    11. Waterton JC,
    12. Robinson SP
    : Oxygen-enhanced MRI accurately identifies, quantifies, and maps tumor hypoxia in preclinical cancer models. Cancer Res 76(4): 787-795, 2016. DOI: 10.1158/0008-5472.CAN-15-2062
    OpenUrlAbstract/FREE Full Text
  40. ↵
    1. Leech M,
    2. Leijenaar RT,
    3. Hompland T,
    4. Gaffney J,
    5. Lyng H,
    6. Marignol L
    : Exploring hypoxia in prostate cancer with T2-weighted magnetic resonance imaging radiomics and pimonidazole scoring. Anticancer Res 43(1): 351-357, 2023. DOI: 10.21873/anticanres.16170
    OpenUrlAbstract/FREE Full Text
  41. ↵
    1. Bar-Zion A,
    2. Yin M,
    3. Adam D,
    4. Foster FS
    : Functional flow patterns and static blood pooling in tumors revealed by combined contrast-enhanced ultrasound and photoacoustic imaging. Cancer Res 76(15): 4320-4331, 2016. DOI: 10.1158/0008-5472.CAN-16-0376
    OpenUrlAbstract/FREE Full Text
  42. ↵
    1. Capkova L,
    2. Koubkova L,
    3. Kodet R
    : Expression of carbonic anhydrase IX (CAIX) in malignant mesothelioma. An immunohistochemical and immunocytochemical study. Neoplasma 61(02): 161-169, 2014. DOI: 10.4149/neo_2014_031
    OpenUrlCrossRefPubMed
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Reprints and Permissions
Correlation of Preclinical In Vivo Imaging Modalities and Immunohistochemistry for Tumor Hypoxia and Vasculature
REBECCA A. D’ALONZO, SYNAT KEAM, TRACY S. HOANG, SUKI GILL, PEJMAN ROWSHANFARZAD, ANNA K. NOWAK, ALISTAIR M. COOK, MARTIN A. EBERT
In Vivo Jan 2025, 39 (1) 55-79; DOI: 10.21873/invivo.13804
Twitter logo Facebook logo Mendeley logo

Jump to section

Keywords