The effect of mild temperature hyperthermia on tumour hypoxia and blood perfusion: relevance for radiotherapy, vascular targeting and imaging

Abstract

Clinically achievable mild temperature local hyperthermia (<43°C) has been demonstrated to be an effective adjuvant to radiotherapy in pre-clinical and clinical studies. In this article, we briefly review the recent progress in the following areas: (1) the effect of mild temperature hyperthermia (MTH) on tumour hypoxia and blood perfusion as assessed by dual marker immunohistochemistry (IHC); (2) the kinetics of MTH induced changes in tumour hypoxia; (3) the potential role of heat-induced tumour reoxygenation on radio- and chemo-sensitisation; (4) the potential role of MTH in combination with vascular targeting agents (VTAs) on tumour response; and (5) non-invasive detection of changes in tumour oxygenation and blood perfusion. It is shown that MTH, by itself or in combination with VTAs, leads to changes in tumour perfusion and oxygenation with potential for radio- and chemo-sensitisation.

• hyperthermia
• hypoxia
• blood perfusion

Introduction

Hyperthermia at temperatures typically obtained in the clinic (<43°C), has been shown to be an effective adjuvant treatment for radiotherapy, improving local tumour control and patient survival in several randomised clinical trials [1–4]. Mild temperature hyperthermia (MTH) has been recommended as part of clinical treatment for advanced cervical cancer patients in combination with radiotherapy; this was based on the Dutch Deep Hyperthermia Trial completed in 1996 and the five trials conducted in Asia, as extensively reviewed by Van der Zee and Van Rhoon [5]. The effectiveness of MTH on radiosensitisation is believed to be due to the dilation of tumour blood vessels and increased tumour oxygenation, as has been shown in many clinical and experimental studies using direct pO₂ measurements [6–9]. Recently, we investigated the effect of heat on tumour oxygenation and blood perfusion at microscopic level using dual hypoxia marker IHC [10]. Interestingly, the effect of MTH on tumour oxygenation and blood perfusion is not homogeneous, and is dependent on the local tumour microenvironment before the treatment, perhaps due to heterogeneity of the tumour microvasculatures. In this article we will briefly review the change of perfusion and hypoxia that occur during and after 40–43°C MTH, and discuss how this effect can be exploited to improve tumour response to the use of hyperthermia as adjuvant therapy.

Effect of MTH on tumour hypoxia and blood perfusion at microscopic level as assessed by dual hypoxia marker IHC

To investigate the effect of MTH on tumour hypoxia at the microscopic level, we used the dual hypoxia marker technique to label the hypoxic cells at different time points before, during, or after heating (41°C). In addition, the observed changes were spatially correlated with the distribution of blood vessels and perfusion as determined by CD31 and Hoechst 33342 staining, respectively.

As Figure 1A shows, the spatial distribution of hypoxic cells as identified by the first hypoxic cell marker pimonidazole and by the second marker EF5 is almost identical (yellow: co-localisation between red, EF5, and green, pimonidazole) in the control tumour (left panel). This suggests that there is no change in tumour hypoxia during the 1.5 h interval between the administering of the two markers. However, significant mismatch between pimonidazole and EF5 was noticed in the heated tumour (right panel). Importantly, in the same tumour section some regions exhibited decreased hypoxia (Region 1, green, i.e., the first hypoxia marker), and other regions increased hypoxia (Region 2, red, i.e., the second hypoxia marker) were observed.

Figure 1. Direct heating effect of mild hyperthermia (41°C) on tumour hypoxia and blood perfusion. (A) Representative composite binary images from control (left) and heated (right) tumours. The first hypoxia marker pimonidazole given 1 h before heat treatment is stained green; the second hypoxia marker EF5 administrated after 30 min heating is stained red; the interval between two hypoxia markers injection is 1.5 h; the overlay of two hypoxia markers (i.e. no change in the hypoxic status), yellow; vasculature endothelium CD31, white; and perfusion marker Hoechst 33342 is blue. The mice were euthanised 1 h after the injection of the second hypoxia marker EF5 and 1 min after the injection of Hoechst 33342. Original magnification 50×, Scale bar = 1 mm. (B) Comparison of binary image detail from region 1 (upper panels, a to d) and 2 (lower panels, e to h) of the heated tumour shown in Figure 1A. From left to right, there are (a and e) pimonidazole, (b and f) EF5, (c and g) CD31 and Hoechst 33342 together, and (d and h) the composite binary image showing all the markers. The upper panel shows a region where hypoxia decreased during heating; the lower panel shows a region where hypoxia increased during heating. Original magnification 50×, Scale bar = 200 µm. (C) Quantitative analysis of the effect of MTH on tumour hypoxia in relation to blood perfusion. In total we analysed 157 ROIs from three tumours that received EF5 after 30 min of MTH, and 127 ROIs from three unheated control tumours injected with two hypoxia markers 1.5 h apart. Each point represents one 1 × 1 mm region of interest (ROI). PF_ROI is defined as the perfusion marker Hoechst 33342 positive fraction in each 1 × 1 mm ROI. HF_ROI is defined as the hypoxia marker pimonidazole or EF5 positive fraction in each ROI. (Redrawn from [10] and unpublished data).

Figure 1B shows a magnified view of regions 1 and 2 from the binary image of Figure 1A. Upon closer examination of the data, the regional differences of the changes in the tumour hypoxia during MTH appear to be related to the functional status of the micro-vasculatures. Specifically, mild hyperthermia decreased hypoxia in the regions with relatively well-perfused blood vessels, but increased hypoxia in regions that were poorly perfused. In Figure 1B, comparing the green and red areas in the left two panels of the top row (Figure 1B, a and b), it can be concluded that MTH reduced tumour hypoxia. This is also evident in the composite image on the right (Figure 1B, d), which shows lesser area of overlap between red and green (yellow) relative to the pre-MTH pimonidazole-positive (green) area. In the third panel from the left (Figure 1B, c), the co-localisation of the perfusion marker Hoechst 33342 (blue) with blood vessels identified by CD31 staining (white), suggests that these vessels were fully functional in this region. In the composite image, the locations of the yellow regions are at a greater distance from the functioning vessels than the green regions. The data in the bottom row of Figure 1B(e–h) are in sharp contrast with that in the top row. First, tumour hypoxia is increased by the treatment (compare the green and red areas in the left two panels (Figure 1B, e and f), and the red and yellow areas in the composite image on the right (Figure 1B, h). Second, in the third panel from the left (Figure 1B, g), there is little Hoechst 33342 (blue) surrounding the CD31 staining (white), suggesting that the blood vessels are dysfunctional in this region (indicated by white arrows). In the composite image, the EF5-stained (red) areas surround the yellow areas, suggesting the development of additional hypoxia around these non-functioning vessels (arrows) upon the application of MTH.

We further quantitatively analysed the image data in 1 × 1 mm ROIs (region of interest) to derive values of perfusion fraction (PF_ROI), pre-treatment hypoxia fraction (pre-treatment HF_ROI, pimonidazole staining positive) and post-treatment hypoxia fraction (post-treatment HF_ROI, EF5 staining positive), and generated scatter plots for these parameters. The left panel in Figure 1C shows that in the control tumour the ratio of post-treatment HF_ROI to pre-treatment HF_ROI of each ROI clustered around 1, and did not vary with the changes in PF_ROI value. In contrast, for the heated tumour (right panel), a negative correlation was observed between the perfused fraction and the ratio of post-treatment HF_ROI to pre-treatment HF_ROI.

Taken together, the heat-induced changes in tumour oxygenation correlate with the hyperthermic effect on tumour blood flow. The most interesting observation of our study is that there are both decrease and increase of tumour hypoxia within the heated tumour, and that the decrease occurs in regions with high Hoechst 33342 staining, while the increase is in regions with low Hoechst 33342 staining. This phenomenon was clearly and consistently observed during and at early time points after MTH. While the mechanism for this observation is not clear at the present, it is possible that the direction of MTH-induced changes in hypoxia depends upon the functional status of regional microvasculatures. As has been reported, tumour vasculatures are highly irregular, e.g. dilated, bulged, constricted, twisted and sharply bent, and mostly lacking a smooth muscle layer and innervations, resulting in an inability to auto-regulate [11], [12]. Normal tissue blood vessels (mainly arterioles) which are incorporated when the tumour invades the normal tissue, are fully capable of responding to external stimuli [13]. Therefore, the regional differences of abnormal tumour vasculatures and normal tissue arterioles might influence the effect of MTH in modulating tumour hypoxia. Consistent with our findings, Ljungkvist et al. in their dual marker studies also found that carbogen breathing caused a drastic reduction in tumour hypoxia that was most marked adjacent to well-perfused regions, and less pronounced in areas abutting the necrotic tumour [14]. All these phenomena indicate the potential influence of the functional status of regional vasculatures on the manipulation of tumour oxygenation.

Kinetics of mild temperature hyperthermia induced changes in tumour hypoxia

The effect of MTH on tumour hypoxia was similar during heating and immediately afterward, but less so 1 h later (Figure 2A). Although MTH induced both increase and decrease in tumour hypoxia in different regions of the tumour, the overall effect was improved oxygenation, an effect which decayed with time. The kinetics of these changes is summarised in Figure 2B left panel (solid bars) with time-matched control tumour (open bars). The greatest reduction in hypoxic fraction was observed immediately after 45 min heating, with the ratio of post-treatment to pre-treatment hypoxia decreasing from 0.95 ± 0.04 to 0.83 ± 0.02 (p = 0.0299); and 1 h afterwards the decrease was from 1.0 ± 0.03 to 0.89 ± 0.03 (p = 0.0322). However, at 6 and 24 h post-treatment, there were no statistically significant changes between the heated and control tumours.

Figure 2. The kinetics of the effect of MTH on tumour hypoxia. (A) The changes in tumour hypoxia and blood perfusion at microscopic level after heating (41°C, 45 min). The colour conventions are the same as Figure 1A. The interval between the second hypoxia marker EF5 injection and heating is given above each image. Original magnification 50×, Scale bar = 1 mm. (B) Xenograft model: Representative kinetics of tumour hypoxia in a different xenograft model as assessed by dual hypoxia marker immunohistochemistry technique (left) or direct pO₂ measurements. Left panel, the ratios of post-treatment HF/pre-treatment HF at different times after heating are shown. Open and solid bars represent unheated and heated tumours, respectively. All the values are means ± SD from 3 to 6 mice/group. Right panel, median pO₂ as a function of time after heating at 41.5°C for 60 min in FSaII tumours and after heating at 41.5°C for 30 or 60 min in SCK tumours are shown. Each data point represents the mean of 10–20 individual tumours with 1 SE shown where larger than the symbol. (C) Spontaneous model: oxygen measurements during a course of fractionated thermoradiotherapy in dogs where 40 CEM43°CT₉₀ were prescribed. Tumour volume and total hyperthermia dose administered are given at the top of each panel. Filled squares are the oxygen measurements. The filled triangles along the abscissa represent hyperthermia treatments while the Xs represent radiation fractions. Fractional hyperthermia doses are given at the bottom of each panel. The line connecting the points is for visual reference only and does not imply that oxygenation status is known between measurements. (Reprinted, with permission, from references [10], [15] and [16]).

Our results are in good agreement with previous experimental and clinical findings which showed that MTH could increase the overall tumour pO₂ level in rodent [6], canine [7] and human tumours [8], [9]. MTH usually improves tumour oxygenation, an effect that occurs rapidly after the initiation of heating. However, the duration of this effect after hyperthermia is both tumour-specific and dose-dependent. Okajima et al. [15] found that the time to restoration to baseline oxygenation after MTH varied with the type of tumour and the heat dose. In SCK mammary carcinoma, tumour pO₂ rapidly decreased to the preheating level within 1 h after 41.5°C for 60 min, but remained elevated for 3 h following 30 min heating. However, in FSaII fibrosarcoma, the median pO₂ remained significantly elevated for 24–48 h after heating for 60 min (Figure 2B, right panel). Thrall et al. [16] investigated the pO₂ changes in spontaneous canine soft tissue sarcomas (n = 7) treated with a fractionated course of radiation and hyperthermia. Two distinct patterns of change were identified in tumour oxygenation following hyperthermia. Hyperthermia improved tumour oxygenation, an effect that persisted throughout the course of fractionated radiation therapy in some tumours (Figure 2C, left panel); but caused a fluctuation or a permanent decrease in tumour oxygenation in other tumours (Figure 2C, right panel).

The above discussion serves to demonstrate that the effect of MTH on tumour pO₂ is complex and depends on multiple factors. Importantly, to devise the optimal use of MTH as an adjuvant to radiation and chemotherapy, more studies are needed to address issues such as (1) the mechanism underlying the kinetic of heating response of different tumour types and whether it is related to their different blood perfusion pattern (well perfused or poorly perfused tumour type); and (2) the development of methods to monitor or predict the response to MTH (e.g. pO₂ measurements, multiparametric IHC analysis, and non-invasive imaging (as discussed in a later section).

Heat-induced tumour reoxygenation and radio- and chemo-sensitisation in the clinic

Tumour hypoxia has long been associated with radioresistance, and also chemoresistance. As discussed previously, our studies at the microscopic level and other studies using direct pO₂ measurements have shown that MTH increases the overall pO₂ level. Thus, its effect on tumour oxygenation may be a beneficial factor when hyperthermia is combined with radiation and/or chemotherapy in the clinic.

In 1996, Brizel et al. [8] reported the effects of radiotherapy and hyperthermia (HT) on soft tissue sarcoma oxygenation and the relationship between treatment-induced changes in oxygenation (based on pO₂ measurement) and treatment outcome. Their study demonstrated that the combination of 10–16 Gy of radiation and hyperthermia significantly increased the oxygenation of human soft tissue sarcoma, but that radiation alone did not improve oxygenation to the same extent. This finding provided the first evidence that hyperthermia may improve tumour oxygenation in human tumours. More importantly, they also found that the magnitude of improvements in tumour oxygenation after the first HT fraction relative to the pretreatment baseline was positively correlated with the amount of necrosis seen in the resection specimen (Figure 3A). More recently, a phase II study by Dewhirst et al. [17] examined the relationship between pO₂, P-31 MRS parameters and the treatment outcome in patients with high-grade soft tissue sarcomas. Their results showed that pO₂ was correlated with pathologic CR ratios (pCR), but the changes in pO₂ after the first heat treatment were not significantly related to the pCR. The pre-treatment average median pO₂ in the first report [8] was 6 mmHg compared with 10 mmHg in this later studies [17], i.e., the tumours in the later series were better oxygenated, which may explain why the reoxygenation was less important in predicting pCR in this report.

Figure 3. Heat-induced tumour reoxygenation and its radio- and chemo-sensitisation in the clinic. (A) The change in tumour pO₂ after HT relative to pretreatment baseline correlated with the extent of necrosis in the resection specimen. The data for each tumour has been normalised to allow for the differences in pretreatment baseline oxygenation (r = 0.55; P = 0.009). (B) Improvement in tumour oxygenation of 11 initially hypoxic locally advanced breast cancers after neoadjuvant chemotherapy and hyperthermia treatment. (C) Changes in average median oxygenation (pO₂) after hyperthermia treatment in responders (CR, complete response; PR, partial response) and nonresponders (NR). Patients who responded to treatment are those who experienced an increase in tumour pO₂ after the treatment. (Reprinted, with permission, from references [8], [9] and [18]).

Changes of oxygenation (pO₂) were also evaluated by the researchers at Duke University in patients with locally advanced breast cancer [9]. Among 18 patients who received neoadjuvant chemotherapy and hyperthermia treatments, tumour hypoxia was present in 11 patients (average median pO₂ = 3.2 mmHg) before treatments. After treatment, median pO₂ increased to 19.2 mmHg (Figure 3B) [9]. In another series of patients treated with concurrent chemotherapy, radiation and hyperthermia followed by mastectomy, those with well-oxygenated tumours before treatments as well as those with significant reoxygenation (with an increase in pO₂ of 15.8 ± 6.6 mmHg) had a more favorable response (CR or PR) to the treatments. However, a significant decrease in pO₂ of 8.3 ± 6.9 mmHg was observed in patients who did not respond to the treatments (Figure 3C) [18].

Although these studies are limited by the relatively small sample size, they do provide important information on the relationship between the changes in tumour hypoxia and response to treatments when hyperthermia is used in combination with radiation and/or chemotherapy. Of course, as heat treatments were given in conjunction with radiation or chemotherapy, hyperthermia is not the sole cause of improved oxygenation. Another possibility is that heat-induced oxygenation may affect the response to the subsequent radiation treatment, or even that of the next several radiation exposures (i.e., 24–48 h or longer after heating).

Tumour response to MTH used in combination with vascular targeting agents (VTA)

Tumour blood flow/perfusion may play an important role in the tumour response to MTH. In general, the poorer the blood supply, the easier it is to heat the tissue and also the better the heat effect; therefore, VTAs may sensitise the tumour vasculature to relatively mild heating. Early studies using hydralazine showed that a reduced blood flow leads to better tumour heating [19], [20]. Numerous preclinical studies (extensively reviewed by Horsman [21]) have demonstrated the benefits of combining heat treatment with VTAs, including angiogenesis inhibiting agents (AIAs) or vascular disrupting agents (VDAs), to improve tumour response. For example, when VDAs (combretastain A-4 disodium phosphate, CA4DP i.p. 30 min before heating; flavone acetic acid, FAA i.p. 3 h before heating; or 5,6-dimethylxanthenone-4-acetic acid, DMXAA i.p. 6 h before heating) were given in combination with MTH (41.5°C with various heating times), the slope of the heat time-response curve increased from 0.02 (heat alone) to 0.06 (heat + CA4DP), 0.08 (heat + FAA) and 0.09 (heat + DMXAA), respectively (Figure 4A) [22].

Figure 4. Preclinical data of MTH in combination with VTAs on tumour response. (A) The effect of VTAs on the response of the C3H mammary carcinoma to heating for various times at 41.5°C. Treatments were heat alone -○-; CA4DP (25 mg/kg i.p. 30 min before heating -▴-); FAA (150 mg/kg i.p. 3 h before heating -▪-); DMXAA (20 mg/kg i.p. 6 h before heating -•-). Results show means (±1 SE) for an average of 14 mice/group with the lines drawn after regression analysis using the individual tumour growth times obtained. (B) Effect of DMXAA and heating on the radiation response of a C3H mammary carcinoma. Tumours were locally irradiated with graded radiation doses and the percentage of mice in each treatment group showing local tumour control 90 days after treatment recorded. Mice were given radiation alone -○-; local heating (41.5°C; 60 min) starting 4 h after irradiation -•-; DMXAA (20 mg/kg) i.p. 1 h after irradiating -▵-; or treated with radiation, DMXAA and heat -▴-. Points are from an average of 13 mice/group. (Reprinted, with permission, from references [21], [22] and [23]).

VTAs have also been shown to be effective at enhancing radiation response in animal tumours. Therefore, the combination of hyperthermia, radiation and VTAs seems to be a logical approach. For example, DMXAA has been shown to enhance thermoradiosensitisation. In C3H mammary carcinoma, combining DMXAA (20 mg/kg i.p. 1 h after irradiation) and MTH (41.5°C, 60 min starting 4 h after irradiation) reduced the TCD₅₀ to 30 (26–35) Gy, relative to 53 (51–55) Gy for radiation alone, 47 (42–52) Gy for DMXAA plus radiation, and 47 (44–51) Gy for heating plus radiation (Figure 4B) [21], [23]. When the heating temperature was decreased to 40.5°C, the effect was decreased but still significant compared with radiation plus DMXAA, or radiation plus hyperthermia. This suggests that if future clinical trials with hyperthermia and radiation were to include a VDA in the treatment regime, the problem of ineffective heating of tumours may be overcome leading to beneficial results [21].

Non-invasive detection of changes in tumour oxygenation and blood perfusion

The benefit of hyperthermia in multimodality cancer treatments are at least partially caused by its modification of the tumour microenvironment, such as changes in perfusion and improvement of oxygenation. Therefore, it is of interest to consider the use of non-invasive imaging techniques of perfusion and hypoxia in the application of thermal therapy. However, there are only a few studies in this area.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a well-established method for assessing perfusion, and is potentially useful for monitoring changes in perfusion due to hyperthermia treatment [24]. Recently, Viglianti et al. [25] reported that DCE-MRI parameters obtained from canine spontaneous soft tissue sarcomas before thermoradiotherapy were predictive of treatment responses. Increases in the signal enhancement curve (AUC) and wash-in rate after the first heat treatment were associated with prolonged metastasis-free and overall survival.

Currently several imaging modalities are being developed that could provide information of tumour oxygenation status. One approach is positron emission tomography (PET) imaging using various hypoxia-targeted radiotracers, such as F-18 FMISO, I-124 IAZGP, and Cu-64 ATSM. Myerson et al. [26] reported the utility of Cu-64 ATSM PET scanning to detect the impact of hyperthermia on tumour physiology. Their results showed that the uptake of Cu-64 ATSM became significantly less in the heated tumours relative to that in control tumours. There is also a good correlation between the increase of pO₂ and the decrease of Cu-64 ATSM uptake.

Conclusion

MTH increases tumour perfusion and this could be utilised to improve drug delivery. At the same time, the increase in tumour oxygenation is expected to increase radiosensitivity, as indicated by numerous preclinical studies [27–29]. However, our studies at the microscopic level show that MTH induced changes of perfusion and hypoxia is heterogeneous within a tumour. The radiobiological consequences due to regional differences on the changes in tumour hypoxia are as yet unknown, and further studies are warranted. Hyperthermia may offer a strategy to modify tumour blood flow and oxygenation leading to improved treatment response. However, this effect in the clinical setting is dependent on thermal dose and tumour type, and additional investigation is needed, especially at the microscopic level, and also with non-invasive imaging techniques.

Acknowledgements

This work is supported by research grants from the National Institute of Health (CA56909, CA109772, and PO1CA115675).

Declaration of interest: The authors state that there are no actual or potential conflicts of interest regarding the contents of this manuscript.