Pathophysiological and vascular characteristics of tumours and their importance for hyperthermia: Heterogeneity is the key issue

Abstract

Tumour blood flow before and during clinically relevant mild hyperthermia exhibits pronounced heterogeneity. Flow changes upon heating are not predictable and are both spatially and temporally highly variable. Flow increases may result in improved heat dissipation to the extent that therapeutically relevant tissue temperatures may not be achieved. This holds especially true for tumours or tumour regions in which flow rates are substantially higher than in the surrounding normal tissues.

Changes in tumour oxygenation tend to reflect alterations in blood flow upon hyperthermia. An initial improvement in the oxygenation status, followed by a return to baseline levels (or even a drop to below baseline at high thermal doses) has been reported for some tumours, whereas a predictable and universal occurrence of sustained increases in O₂ tensions upon mild hyperthermia is questionable and still needs to be verified in the clinical setting. Clarification of the pathogenetic mechanisms behind possible sustained increases is mandatory.

High-dose hyperthermia leads to a decrease in the extracellular and intracellular pH and a deterioration of the energy status, both of which are known to be parameters capable of acting as direct sensitisers and thus pivotal factors in hyperthermia treatment. The role of the tumour microcirculatory function, hypoxia, acidosis and energy status is complex and is further complicated by a pronounced heterogeneity. These latter aspects require additional critical evaluation in clinically relevant tumour models in order for their impact on the response to heat to be clarified.

• tumour blood flow
• tumour oxygenation
• tumour pH
• tumour bioenergetics
• hyperthermia

Introduction

Several pathophysiological features of the tumour microenvironment greatly affect the efficacy of anticancer treatments, hyperthermia (HT) included [1], [2]. Amongst these factors, tumour blood flow (TBF) plays a seminal role since (1) it critically determines convective heat dissipation and thus the ability to heat the tissue, and (2) it greatly influences the tumour microenvironment and subsequently the sensitivity of the tumour to heat. A compromised tumour blood flow results in the development of adverse microenvironmental conditions (e.g. hypoxia, acidosis, energy depletion [3]). Cells living under such hostile conditions seem to be more sensitive to therapeutic hyperthermia. TBF and the tumour microenvironment thus play a significant role in determining the response of tumour cells to heat as evidenced in experimental (preclinical) as well as clinical settings. Besides sensitising cancer cells to heat, pathophysiological characteristics of the tumour microenvironment are themselves affected and altered by tissue heating in return, thus creating a complex (and partly self-perpetuating) situation, depending on the direction, extent and duration of individual changes upon heating.

These changes in microenvironmental factors upon heating have been described in detail previously, especially for experimental tumours in the preclinical setting [4–8]. Comparable studies on human tumours are very scarce because most of the data are derived from anecdotal observations [4]. In addition, in many of the investigations, hyperthermia at temperatures above 42.5°C have been used, a situation which is only rarely achievable in the clinical setting, where–in most cases–hyperthermic tissue temperatures of between 39°C and 42.5°C have been reported [9], [10]. Whereas in the preclinical HT studies on experimental rodent tumours tissue temperatures were relatively uniform, in human tumours spatially heterogeneous temperature fields have often been described [4].

In this review, updated data on relevant parameters of the tumour micro-milieu are presented with special emphasis on changes occurring during ‘mild’ or ‘low dose’ hyperthermia, i.e. within the temperature range 39–42.5°C. To provide a basis for this review, a general description of the pathophysiological tumour microenvironment is presented first. Furthermore, particular emphasis is given to the in vivo situation, which is rather more complex than that of in vitro tumour biology.

Pathophysiological characteristics of the tumour microenvironment

The physiology of tumours is uniquely different from that of normal tissues. It is characterised by O₂ depletion (hypoxia and anoxia), extracellular acidosis, high lactate levels, glucose deprivation, energy impoverishment, significant interstitial fluid flow and interstitial hypertension [11–22]. The O₂ depletion found in tumours is not the result of increased oxygen consumption rates, but instead is due to an imbalance between perfusion and consumption, so that changes in both of these factors must be taken into consideration. Since the hostile environment is largely determined by an abnormal tumour microcirculation (Table I), the features of the tumour microcirculation and changes upon hyperthermia will be discussed first. (The term microcirculation is generally used to describe the functions and structures of microvessels with outer diameters <100 µm, terminal lymphatics included.)

Table I.  Factors characterising the hostile pathophysiological tumour microenvironment (‘crucial Ps’).

A. Increase in	B. Decrease in
Perfusion inadequacies and vascular chaos	Partial pressure of O2 (pO2)
Perfusion heterogeneities	Production of high-energy compounds
Permeability in tumour microvessels	pH of extracellular compartment
Pressure of interstitial fluid	Perfusion pressure (arterio-venous pressure difference in microvessels)
Production of lactate
Production of adenosine
Paucity of nutrients
Partial pressure of CO2 (pCO2)
Paucity of bicarbonate

All listed parameters change for the worse during tumour growth and during tumour heating. Modified from Vaupel [3].

Tumour angiogenesis, vascularity and blood flow

When considering the continuous and indiscriminate formation of the vascular network in a growing tumour, five different pathogenetic mechanisms can be discussed: (1) angiogenesis by endothelial sprouting from pre-existing venules, (2) co-option of existing vessels which can keep their normal structure and function, (3) vasculogenesis (de novo vessel formation through incorporation (recruitment) of circulating endothelial precursor cells from bone marrow), (4) intussusception (splitting of the lumen of a vessel into two), and (5) formation of pseudovascular channels lined by tumour cells rather than endothelial cells (‘vascular mimicry’ [3]). The tumour vasculature is characterised by vigorous proliferation leading to immature, structurally defective, and–in terms of perfusion–ineffective microvessels. Consequently, tumour blood flow is chaotic and heterogeneous, both spatially and temporally (‘4D-heterogeneity’ of TBF).

In contrast to co-opted vessels, newly formed microvessels in most solid tumours do not conform to the normal morphology of the host tissue vasculature. Microvessels in solid tumours exhibit a series of severe structural and functional abnormalities. They are often dilated, tortuous (convoluted), elongated, and saccular. There is significant arterio-venous shunt perfusion accompanied by a chaotic vascular organisation that lacks any regulation matched to the metabolic demands or functional status of the tissue. Excessive branching is a common finding, often coinciding with blind vascular endings. Incomplete or even missing endothelial lining, lack of pericytes and interrupted (discontinuous) basement membranes result in an increased vascular permeability with extravasation of blood plasma and of red blood cells expanding the interstitial fluid space and drastically increasing the hydrostatic pressure in the tumour interstitium (interstitial fluid pressure). In solid tumours there is a rise in viscous resistance to flow caused mainly by haemoconcentration. Reductions in red blood cell flexibility brought about by vascular hypoxia, low pH or hyperthermia can also contribute to the increased viscous resistance [23], [24]. Aberrant vascular morphology and a decrease in vessel density are responsible for an increase in geometric resistance to flow, which can lead to an inadequate perfusion.

Substantial spatial heterogeneity in the distribution of tumour vessels and significant temporal heterogeneity in the microcirculation within a tumour [3] may result in a considerably anisotropic distribution of tumour tissue oxygenation and of a number of other factors which are usually closely linked and which define the so-called metabolic microenvironment. Variations in these parameters between tumours are often more pronounced than differences occurring between different locations or microareas within a tumour [3].

Blood flow can vary considerably, ranging from 0.01 to 2.9 mL/g/min, both in the preclinical and clinical setting [3]. Tumours can thus have flow rates similar to those measured in tissues with a high metabolic rate or can exhibit perfusion rates comparable to those of tissues with a low metabolic turnover. Flow data from multiple sites of measurement show marked heterogeneity within individual tumours. However, tumour-to-tumour variability seems to be more pronounced than intratumour heterogeneity.

Angiogenesis, microcirculation and tumour blood flow upon hyperthermia

Hyperthermia can inhibit angiogenesis, a mechanism which has been suggested to play a role in tumour regression caused by HT [25]. These authors reported that angiogenesis inhibition occurs as a result of direct injury to endothelial cells (EC). An inverse relationship was demonstrated between vessel growth and treatment temperature over the range of 41–44°C after 30 min hyperthermia treatment [26]. Following these earlier experiments, Sawaji et al. [27] were able to show that the anti-angiogenic action of HT may be due to the down-regulation of the expression of tumour-derived vascular endothelial growth factor (VEGF) production, thereby inhibiting proliferation of ECs and extracellular matrix remodelling in blood vessels. In addition, HT can activate a specific gene response that involves the transcription of the human plasminogen activator inhibitor-1 (PAI-1), the key regulator of the plasminogen activation pathway [28]. The latter study indicated that the heat-mediated PAI-1 induction may be an important pathway by which HT exerts its antitumour activity.

Most studies on HT-induced changes in tumour microcirculation date back to the early 1980s [29–43]. Summarising these observations, it can be stated that the effect of heat on tumour microcirculation is temperature- and time-dependent. Whereas mild HT can (temporarily) increase microcirculatory functions, higher temperature and/or longer exposure times can impair the microcirculation and damage the tumour microvasculature. Normal vasculature responds differently to the fragile and chaotic tumour microvasculature with microvasculatory flow increasing as temperature increases even at temperatures that produce a complete shutdown of tumour microvasculature (T > 42.5°C).

The immediate physiological reaction after elevation of tissue temperature is a change in blood flow [44]. The alteration in perfusion upon heating is in general much greater in normal tissue than in tumours. Normal tissues often included into the heating volume are skeletal muscle and skin with flow values of 0.04 mL/g/min (resting state), and 0.08 mL/g/min (under thermoneutral conditions), respectively [45]. Skeletal muscle and skin can increase flow by at least a factor of 15 whereas tumour perfusion may only increase 2-fold [4], starting from a baseline which is often significantly higher than flow values observed in skeletal muscle and skin.

Changes in tumour blood flow upon hyperthermia in experimental tumours have been comprehensively reviewed [4], [39]. On average, blood flow in experimental tumours temporarily increased upon heating and declined thereafter. The extent and time course of the heat-induced initial flow increase depends on the heating-up rate, hyperthermia level, heating time, temperature homogeneity achieved, tumour type, and growth site of the experimental tumour. Not surprisingly, heat-induced flow changes also depend on the absolute blood flow rate of the tumours before treatment. In experimental tumour models, heat-induced changes in blood flow were also seen to be related to tumour volume, with larger tumours exhibiting the lower flow values [4], [5], [39]. Temperature- and time-dependent changes in tumour perfusion have been reported in a large series of isografted mice and rat tumours and in xenografted human tumours.

Kelleher et al. [46] measured red blood cell (RBC) fluxes simultaneously in subcutaneous rat tumours at multiple sites using invasive laser Doppler flow probes. From these measurements the following conclusions could be drawn:

1. The response of the tumour to heat was dependent on tumour volume: the smaller the tumour (i.e. the higher the initial blood flow values), the less pronounced (and/or less often) the shutdown of microcirculatory function following a slight, initial increase in flow.

2. The extent and direction of flow changes observed within a tumour did not correlate with the tumour site (peripheral versus central regions) or with the actual tissue temperature at the site of measurement (42.4°C versus 44.5°C).

3. Variability in RBC flux responses to heat was similar in small and large tumours.

4. Whereas RBC flux substantially increased by a factor of 2.3 in some tumour areas during the heating period, in others it steadily decreased by up to 75% upon hyperthermia; biphasic changes in flux could often be observed with an initial flow increase in many cases (see Figure 1).

Figure 1. Variability of tumour perfusion response to localised 44°C hyperthermia. Curves show relative red blood cell flux at three different sites within a single tumour at comparable temperatures. The shaded bar shows the time of hyperthermia application (modified from Vaupel et al. [47]).

In general, there was a clear indication from these preclinical experiments, that–at least at the start of heating and during the early phase of steady-state tissue temperatures–the intratumour variability of blood perfusion increased, both spatially and temporally, indicating an increased ‘4D-heterogeneity’ in tumour perfusion, even when heating is rather uniform. Flow values are finally ‘homogenised’ on a lower level upon shutdown of flow at a later phase or after heating. Heat-induced flow changes seen in experimental tumours are not at all predictable due to pronounced heterogeneities.

At least in experimental tumour models, the heating techniques used may have an additional impact on the biological behaviour. In the case of saline bath hyperthermia (caveat: immersion of a tumour in a hypo-osmolar water bath creates an osmotic water shift leading to oedema formation which adds to the heat-induced oedema caused by an increased vascular permeability at higher temperatures), the highest temperatures are to be found in the tumour shell, which in many experimental systems exhibits the highest flow rates before treatment. In contrast, other heating devices lead to a maximum energy deposition in more central parts of the tumours which are often characterised by lower perfusion rates. Thus tumours showing comparable mean tissue temperatures during hyperthermia may show completely different patterns of maximum heat deposition with respect to high- and low-flow regions [5]. Heating rodent tumours at 40–42°C (i.e. to clinically achievable hyperthermia levels [48]) was found to cause an enduring increase in blood flow (and vascular permeability) thus possibly improving the delivery of chemotherapeutic drugs or immunotherapy agents to tumour cells [49], [50].

In animal tumour models, the response of blood vessels to multiple heating differs markedly from that of single heating [39]. Increases seen in tumour blood flow during a second heating following the ‘conditioning of blood vessels’ by the first heating appear to be due to the development of ‘vascular thermotolerance’, provided that the second heating is applied within 48 hours at relatively mild temperatures (data reviewed by Griffin et al. [51]), indicating a considerable impact of the treatment regime (single versus fractionated, HT alone or in combination) (caveat: the definition ‘vascular thermotolerance’ used by Griffin et al. [51] is not compatible with the traditional definition of Gerner and Schneider [52] and Henle et al. [53]).

Temperature-dependent changes in tumour perfusion have also been reported in spontaneous canine tumours 24 h after hyperthermia treatment with increases in flow upon hyperthermia at lower temperatures, whereas high temperature hyperthermia treatments were more likely to cause vascular damage. In addition, the data suggest that the thermal threshold for vascular damage is higher in the clinical setting than in rapidly growing rodent tumours [54]. In a recent paper by Viglianti and colleagues [55] on canine tumours, changes in perfusion-related parameters after thermoradiotherapy persisted for 24 h after treatment. So far, the underlying mechanisms for this effect are, however, not known. The changes found predicted not only for local tumour control, but also for metastasis-free and overall survival.

Since perfusion is crucial for hyperthermia treatment planning in the clinical setting, assessment of flow changes in human tumours upon hyperthermia are of utmost importance. However, as mentioned above, data from studies on human tumours are still scarce and numerous questions still exist regarding whether or not results obtained from fast-growing murine tumours are an appropriate reflection of what occurs clinically in human tumours, especially when the fact that experimental tumours usually exceed 2% of the animal's body weight is taken into account (i.e. these tumours are considered to be ‘bulky’ tumours [56]).

In a first series of tumour blood flow measurements upon local hyperthermia, Waterman et al. [57], [58] described an increase in tumour blood flow by amounts varying from 15% to 250% during the first 20–50 min of heating at 41–45°C, after which it remained relatively constant during the remainder of the treatment session. A sharp reduction in flow was never observed in this study (caveat: heat transport by conduction was neglected in these measurements [59]). Subsequently, Waterman et al. [60] studied the effects of localised heating on tumour blood flow in human patients using thermal wash-out techniques. In these investigations, great emphasis was laid on the importance of thermal conduction in the dissipation of heat energy during local hyperthermia [59], [61]. In the Waterman study [60], the authors were unable to demonstrate a flow drop during individual hyperthermia sessions of 1 h duration with temperatures as high as 45°C. Similar results were reported by Lammertsma et al. [62] who measured blood flow upon HT in breast cancers using positron emission tomography. A study of lung and breast cancer metastases in the lymph axillary nodes showed an increase in tumour blood flow after each hyperthermia session relative to the pretreatment perfusion during the course of fractionated hyperthermia. In some cases, blood flow rises of up to 200% were found [63]. Temperature profile assessments [64] and PET [65] have been used to observe the effects of heating on blood flow in human pelvic tumours. In these experiments, regional hyperthermia was applied for about 60 min at temperatures of between 40°C and 42°C, but significant changes in perfusion upon hyperthermia were not observed (caveat: in rectal cancers, blood flow varied substantially depending on the methods used to estimate perfusion in rectal cancers: 0.09 mL/g/min with temperature profile assessments [64], and 0.36 mL/g/min with PET [65]).

When the microcirculatory function of human tumours was evaluated using the laser Doppler flow technique, four different temporal patterns were observed during hyperthermia which had been applied 1–2 h after irradiation. These ranged from a steady increase in flow to a plateau level (28%), no change in RBC flow (36%) or a steady drop in flux (23%), to an initial substantial increase in laser Doppler flow followed by a slight decrease during hyperthermia (14% of cases [66]). These different patterns did not correlate with the average temperature recorded at the site of measurement (42–44°C for approximately 57 min).

A lack of changes seen in thermal wash-out rates in rectal carcinomas, soft tissue sarcomas and other tumours are indicative of only minor changes in blood flow during fractionated hyperthermia immediately after radiotherapy [67]. The response of human tumour blood flow to fractionated thermoradiotherapy has also been described by Waterman et al. [68]. In this study, blood flow in human tumours decreased during the fractionated course of radiotherapy. The overall percentage drop of blood flow ranged from 50–100%. The steady-state temperature at which blood flow started to decrease ranged from 40–43°C.

Taken together, there is no clear indication of changes in flow in human tumours to match those described in the experimental setting, i.e. a transient or steady increase in tumour blood flow in the early phase or during localised hyperthermia, and an overall decrease in tumour blood flow in the later phase or after heating at temperatures >42.5°C. In human tumours therefore, blood flow change might not be the key mechanism by which hyperthermia exerts its beneficial clinical effects.

In many reviews on the behaviour of tumour blood flow upon hyperthermia, relative changes in perfusion are shown for normal tissue (skin, muscle) in comparison to experimental tumours [37]. In this type of data presentation, tumour blood flow–on average–goes down, whereas blood flow in representative normal tissues goes up. This way of representing relative flow values can, however, be completely misleading and often results in inappropriate conclusions being made concerning the flow behaviour of human tumours upon heat treatment. When absolute flow values are taken into consideration, a totally different picture becomes apparent [56], [69] which does not support the current concept of HT-induced blood flow changes being used to explain the rationale of HT in the clinical setting. For example, blood flow in breast cancers is, on average, significantly higher than in normal breast tissue (e.g. 0.30 mL/g/min in breast cancer and 0.06 mL/g/min in normal breast tissue [14], [62]), and changes in flow upon hyperthermia do not support the notion of HT-induced flow increases [14], [62] since tumours undergoing HT failed to show any flow changes at all, i.e. upon HT, flow values in breast cancers are consistently higher than in normal breast tissue. Blood flow values in cervix and prostate cancer can–at the least–be three times higher than in normal cervix and prostate, making a quasi selective heating of the malignant tissue very difficult [70], [71].

It is important to recognise that perfusion, with respect to nutrient delivery is quite different from that responsible for heat transfer: Thermally significant vessels are generally ‘macrovessels’ (diameter >100 µm), whereas nutrient transport and exchange generally occur in the microcirculation (vessel diameter <100 µm).

Cellular oxygen consumption at elevated temperatures

In general, the actual uptake of O₂ and nutrients by a tissue is determined by the respective availabilities in the microcirculation, the diffusional flux in the interstitial compartment, and the metabolic requirements of the cells. This holds true for both normal tissue and solid tumours during normothermia and hyperthermia. Under in vitro conditions, however, where no supply limitations are present, the capacity of the cells to consume oxygen is the limiting factor. The in vivo O₂ availability (i.e. microcirculatory function at normoxaemia) is the paramount limiting parameter for many solid tumours. Accordingly, changes in the in vivo O₂ uptake during hyperthermia only occur when tumour heating is accompanied by changes in nutritive blood flow [4]. In other words, O₂ uptake rates in vivo will parallel changes in nutritive flow in either direction.

The temperature dependency of cellular oxygen consumption rates has been reviewed by Vaupel [4]. On average, isolated tumour cells (exponentially) increase their O₂ consumption rate following temperature elevation up to maximum values at 41–42.5°C. DS-sarcoma cells suspended in native ascitic fluid exhibit a maximum O₂ consumption rate of 42 µL/g/min at 42°C (Figure 2). This consumption rate is comparable to that of the brain at 37°C [14].

Figure 2. Temperature dependency of the O₂ consumption rate of isolated ascites cells suspended in native ascites fluid (DS-sarcoma) (modified from Vaupel [4]).

Assuming a temperature rise during hyperthermia from 32° to 42°C (e.g. in a superficial tumour) and taking into account the data presented in Figure 2, a 2.3-fold increase in the respiration rate is observed, (i.e. the van’t Hoff quotient, Q₁₀, is 2.3), indicating an undisturbed respiration activity of the cell line investigated. These data speak strongly against a mitochondrial dysfunction, which is often thought to be a general hallmark in malignant growth [72].

Tumour tissue oxygenation upon hyperthermia

Over the last two decades, numerous studies have clearly shown that most locally advanced solid tumours contain hypoxic tissue areas [e.g. 3, 73] and as such are a characteristic of malignant growth. The pathogenesis of tumour hypoxia is multifactorial (Table II).

Table II.  Chronic and acute tumour hypoxia: Classification according to different causative mechanisms (selection).

A. Acute hypoxia (cyclic, intermittent, fluctuating, transient, repetitive hypoxia)	B. Chronic hypoxia (sustained hypoxia)
1. Ischaemic hypoxia due to perfusion limitations following temporary obstruction of vessels (e.g. by tumour cells, blood cell aggregates, fibrin aggregates)	1. Diffusion-limited hypoxia due to surpassing of the maximum O2 diffusion distances, exaggerated longitudinal intravascular O2 gradients or shunt perfusion
2. Hypoxemic hypoxia due to severe reduction of intravascular O2 content (e.g. transient plasma flow, temporal fluctuations of red	2. Hypoxemic hypoxia due to anaemia (tumour-associated, therapy-induced, HbCO formation)
blood cell flux)	3. Hypoxia due to compromised perfusion through leaky microvessels (e.g. abolished perfusion pressure difference in areas with high interstitial fluid pressure)
	4. Hypoxia due to compromised perfusion as a result of a reduction in red blood cell flexibility (e.g. under hypoxia, low pH and hyperthermia)

Hypoxic and/or anoxic tissue areas are heterogeneously (both temporally and spatially) distributed within the tumour mass. Moreover, hypoxia has been shown to be a significant negative prognostic factor in many clinical studies, independent of the treatment type [74].

In addition, hypoxia can lead to therapeutic resistance to standard radiotherapy, some forms of chemotherapy and photodynamic therapy [1], [75]. In contrast, hypoxia may increase the therapeutic effects of HT, most probably via indirect effects (e.g. acidosis, energy depletion). There is no convincing evidence so far for a direct effect of hypoxia per se on the efficacy of HT [76], although a study by Overgaard and Bichel suggested a slightly enhanced hyperthermic response under hypoxic conditions [77].

Evidence supports the notion that hypoxia is an important driving force in malignant progression, since it can promote the development of a malignant phenotype [1], [3], [78], [79]. Hypoxia–as an inherent consequence of unregulated growth–can further local invasion, intravasation of cancer cells and finally metastatic spread to distant sites whereby this occurs in a cooperative manner at different mechanistic biological/molecular levels [3], [79], [80].

Tumour cell variants with adaptations favourable to survival under hypoxic stress may have growth advantages over non-adapted cells in the hypoxic microenvironment and may subsequently develop a more aggressive phenotype which in turn is responsible for malignant progression. Cyclic hypoxia and pronounced spatio-temporal heterogeneities (4D-heterogeneities) in hypoxia may be the most powerful factors promoting an aggressive tumour phenotype, which drive local spread and distant metastasis [79–85].

Due to this crucial role of tumour hypoxia in the clinical setting, the tumour oxygenation status has attracted the interest of numerous researchers over the last two decades [73]. Since tumour hypoxia has been considered to be an indirect sensitiser for HT, spatial and temporal changes of tumour oxygenation have also been evaluated upon tumour heating.

When the oxygen status is assessed for characterisation of solid tumours at elevated temperatures in the preclinical setting, there is clear evidence that low thermal doses can improve tissue oxygenation provided that tumour perfusion exhibits an increase under these conditions [4], [5]. At higher thermal doses, tissue oxygenation deteriorates severely, reflecting the breakdown of tumour microcirculation under these conditions [40], [89–93]. From these experiments it has been concluded that changes in tissue oxygenation upon HT may directly reflect simultaneous alterations in tissue perfusion.

There is some controversy concerning the duration of the improvement in tissue oxygenation upon mild hyperthermia, with a series of studies reporting a rapid return to a normal oxygenation status in tumours after the cessation of heating, i.e. the effects observed are rather short-lived (Figure 3) [89–93]. In this context it has also been noticed that with higher tissue temperatures the initial increase and the following drop in tissue oxygenation were more pronounced (Figure 4). In contrast to these data, other studies have shown that the improvement in tumour oxygenation can actually last for up to 24–48 h after heating using low-dose HT [38], [49], [94–98].

Figure 3. Tumour pO₂ values measured continuously before, during and after 40°C, 41.8°C or 43°C hyperthermia. Each data point indicates the mean value (±SEM) for the number of tumours indicated. The shaded area indicates the time of heating (modified from Kelleher and Vaupel [93]).

Figure 4. Comparison of mean pO₂ values before, during (at t = 35 min after reaching the set temperature) and after localised hyperthermia. Each data point indicates the mean value (±SEM) for the number of tumours indicated. * = p < 0.01 (modified from Kelleher and Vaupel [93]).

However, long-lasting improvements in the tumour oxygenation status following mild hyperthermia treatments are difficult to explain because the perfusion changes that most probably account for the enhanced tumour oxygenation during mild HT should return to baseline at or soon after the cessation of heating [2]. If the data showing a long-lasting improvement in the oxygenation status could be confirmed by other groups, using different tumour models and other heating techniques, the sustained increase in the oxygenation status could enhance the response of tumours to radiotherapy and O₂-dependent chemotherapy/radiochemotherapy. A sustained increase in blood flow (as suggested by the improvement in tumour oxygenation) could probably enhance delivery of anticancer agents. However, since clinically relevant increases in the tissue oxygenation which endure beyond the heating period are not consistently seen in preclinical studies, irrespective of the tumour temperature level [93], an improvement in the efficacy of O₂-dependent cancer therapy is unlikely to be achieved in the post-hyperthermia period if these contradictory results can be translated to the clinical setting.

There are clinical observations that seem to support the finding of a prolonged improvement in oxygenation at mild HT temperatures. These data were obtained in the clinical setting during fractionated hyperthermia and radiation therapy [10], [54], [99], hyperthermia treatment and chemotherapy [100] or thermochemoradiotherapy [48]. These findings, however are in contrast to those reported by Zywietz et al. [101] who observed a progressive deterioration in the oxygenation status with combined fractionated radiation and hyperthermia (43°C, 1 h) which was followed by a continuous shrinkage of the tumours and most probably a decline in the overall tumour O₂ consumption resulting from heat-induced cell death.

It is unlikely that increased tumour blood flow produced by low thermal doses represents the only mechanism related to improvement of oxygenation. This is supported by the fact that no relationship between changes in perfusion and comparable alterations in tumour oxygenation in the clinical setting [54] and in experimental tumours [92] were seen.

Another likely cause for the improved tumour oxygenation might be a heat-induced decrease in oxygen consumption. However, the respiration rates of tumour cells are not significantly different at 37°C and 43°C (see Figure 2). A maximum uptake of O₂ is to be expected at 42°C. Therefore, it cannot be concluded that a decrease in O₂ consumption together with an increase in tumour blood flow may be the general mechanisms for the improvement of tumour oxygenation after hyperthermia at moderate thermal dose as suggested by Vujaskovic and Song [44]. Further investigations will be necessary if clearer insights into the relationship between thermal parameters, tumour perfusion and oxygenation upon hyperthermia are to be obtained.

Tumour pH upon heat treatment

Warburg's classic work in the 1920s showed that cancer cells intensively convert glucose to lactic acid even in the presence of oxygen. Because of this excessive lactic acid production it was assumed for many decades that tumours are acidic [14], [18], [102–108]. However, the unfolding story of tumour pH and its consequences have become clearer over the last two decades, due to the availability of techniques capable of preferentially measuring intra- or extracellular pH in malignancies. Under many conditions it has now been confirmed that the intracellular pH in tumour cells is neutral to alkaline as long as tumours are not oxygen- and energy-deprived [22], [109]. Tumour cells have efficient mechanisms for exporting protons into the extracellular space, which represents the acidic compartment in tumours [13],[110–112]. For this reason, a pH gradient exists across the cell membrane in tumours (pH_i > pH_e). Interestingly, this gradient is the reverse of that found in normal tissues where pH_i is lower than pH_e [3], [14], [22], [113–116]. In spontaneous canine tumours assessed by Prescott et al., the magnitude of the pH gradient varied widely and individual tumours had both positive and negative pH gradients [117].

As already mentioned, hypoxic cancer cells intensively split glucose to lactic acid. There are no longer any reasons to ascribe the aerobic glycolysis found as being specific to malignant growth although the increased capacity for glycolysis still remains a key feature of tumours. Other relevant pathogenetic mechanisms yielding an intensified tissue acidosis are based on substantial ATP hydrolysis, glutaminolysis, ketogenesis and CO₂/carbonic acid production.

pH values measured with invasive electrodes preferentially reflect the acid-base status of the extracellular space (pH_e) which occupies approximately 45% of the total tissue volume in malignant tumours. Using pH microsensors, a remarkable variability in measured values could be documented which exceeds the heterogeneity observed within tumours. The results of pH measurements using ³¹P-nuclear magnetic resonance spectroscopy yield pH values which mainly reflect the pH found in the cytosol [14], [118].

Reduced pH values (acidosis) can sensitise tumour cells to single heat treatments and to fractionated hyperthermia [119]. Based on this knowledge, a series of measurements has been performed on the impact of localised heating on intra- and extracellular pH values [4], [5], [37], [119]. Summarising the relevant data, it can be stated that tumour temperatures >42.5°C and appropriate heating times (>30 min) can reduce both intracellular and extracellular pH, which may further sensitise tumour cells to hyperthermia in the sense of a positive feedback mechanism.

Relevant pathogenetic mechanisms leading to an intensified acidosis upon heat treatment (which is reversible after HT) are:

1. an increased glycolytic rate with accumulation of lactic acid,

2. an intensified ATP-hydrolysis,

3. an increased ketogenesis with accumulation of acetoacetic acid and β-hydroxybutyric acid,

4. an increase in CO₂ partial pressures,

5. changes in chemical equilibria of the intra- and extracellular buffer systems (ΔpH/ΔT = −0.016 pH units/°C), and

6. an inhibition of the Na⁺/H⁺ antiporter in the cell membrane [5].

The significance of the baseline pH_e and pH_i values and changes occurring during hyperthermia in clinical studies still remains unclear, since–in contrast to in vitro and in vivo data derived from experimental tumours treated with hyperthermia alone–the results are equivocal so far. There are contradictory results in the clinical setting concerning the direction of heat-induced changes in pH (increase versus decrease), the prognostic power of pH_i or pH_e before treatment using HT or thermoradiotherapy, etc. [54], [106], [120–124]. In the study of Lora-Michiels et al. [122], the pH changes seen predicted for metastasis-free and overall survival. Further clarification of these effects is urgently needed.

Bioenergetic status of tumours upon hyperthermia

In recent years the bioenergetic status of solid tumours has been assessed by ³¹P-nuclear magnetic resonance (³¹P-NMR, ³¹P-MRS), high performance liquid chromatography (HPLC) and by quantitative bioluminescence [3], [14]. Consistently found changes are: decreased ATP and PCr concentrations, increased inorganic phosphate (P_i), ADP, AMP and (extracellular) adenosine concentrations and lower PCr/P_i and ATP/P_i ratios, which have been reported in solid tumours as compared to their tissues of origin. For example, ATP content of soft tissue sarcomas was on average 0.5 mM compared to approximately 5 mM in normal skeletal muscle [14]. This situation has been repeatedly described for tumours with median pO₂ values <10 mmHg, mean glucose concentrations <2 mM and mean lactate levels in the tumour tissue >10 mM [3], [125].

Cellular ATP depletion (most probably caused by substrate deprivation) is thought to sensitise tumour cells to hyperthermia [76], [126], [127]. Upon heating at temperatures >42.5°C and heating times >30 min, a further drop in the ATP and PCr concentrations has consistently been reported [4], [5], whereas the levels of P_i, ADP, AMP and adenosine increased further under HT conditions. Accordingly, ATP/P_i and PCr/P_i ratios substantially dropped at high thermal doses which would probably result in a sensitisation of tumours to heat treatment. At 41°C for 15 min, the ATP/P_i ratio only fell by approx. 30% for approximately 6.5 h [128].

The ATP decline observed upon heat treatment is mostly due to

1. an increased ATP turnover rate (i.e. intensified ATP hydrolysis). As a result of an increased ATP degradation, an accumulation of purine catabolites (adenosine included) has to be expected together with a formation of H⁺ ions and reactive oxygen species at several stages during degradation to the final product uric acid [3],

2. a poorer ATP yield (on a molar basis) as a consequence of a shift from oxidative glucose breakdown to glycolysis.

In many human tumours (other than brain tumours), ³¹P-NMR has often shown high concentrations of phosphomonoesters (PME) and phosphodiesters (PDE). The PME signal primarily includes membrane phospholipid precursors. The PDE peak was identified to be largely a result of membrane phospholipid decomposition products. Upon hyperthermia treatment (>42.5°C, ≥30 min) the PME/ATP ratio significantly increased, whereas the PME/P_i and PDE/P_i ratios decreased following heating [129].

Interestingly, PME/P_i and PDE/P_i ratios have been described as being significant predictive factors for treatment outcome upon thermoradiotherapy in addition to the oxygenation status (pathological cure rate), and the PCr/PDE, as well as the PME/PDE ratios (overall survival) [130].

Conclusions

Thermal sensitivity has been shown to greatly depend on the efficacy of tumour blood flow and parameters defining the metabolic microenvironment such as hypoxia, acidosis, substrate deprivation, accumulation of metabolic waste products, and energy depletion. If recent experimental data are critically evaluated, there is some evidence that besides microcirculatory function, intracellular pH and the bioenergetic status may be decisive factors ultimately modulating the thermosensitivity of cancer cells.

Results obtained from patient malignancies (canine, human) treated with adjuvant mild hyperthermia (T < 42°C) suggest that therapeutically relevant changes in these microenvironmental parameters may be quite different from those seen in fast-growing, low-flow rodent tumours upon heat treatment. Thus, biological principles that seem favourable in experimental rodent tumours may not necessarily hold for tumours (or tumour areas) with high perfusion rates. In addition, the pronounced heterogeneity between tumours and within tumours may lead to a non-uniform response to heat treatment, especially when hyperthermia is used as an adjuvant to established treatment modalities (radio- and/or chemotherapy).

Acknowledgements

The valuable editorial assistance of Anne Deutschmann-Fleck in preparing this manuscript is greatly appreciated.

Declaration of interest: Both authors gratefully acknowledge project grants from the Erwin Braun Foundation, Basel (Switzerland). The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.