The Clinical Importance of Assessing Tumor Hypoxia: Relationship of Tumor Hypoxia to Prognosis and Therapeutic Opportunities

To the Editor:

We have read the comprehensive invited review of J.C. Walsh et al. in this Journal with great interest (20). The authors describe the clinical importance of assessing tumor hypoxia and its relationship to prognosis and therapeutic opportunities. We believe that this information is of utmost interest to experimental and clinical oncologists since tumor hypoxia is a hallmark feature of most locally advanced patient tumors, which promotes tumor progression and adversely impacts on the efficacy of different anticancer therapies (13, 14, 16, 17). Pretherapeutic hypoxia has been identified as a strong and independent adverse prognostic factor for patient outcome [e.g., hypoxic cancers of the uterine cervix, soft tissue sarcomas, and head and neck cancers (15)].

Our institution has contributed key information to the field of tumor hypoxia and its consequences in the clinical setting, which are partly referenced in the article of Dr. Walsh et al. However, we feel that this review may benefit from some additional information that might be of interest for the distinguished readership of this highly reputed journal.

In Section II.A (Pathophysiology of Hypoxia), the authors state that tumor hypoxia can be the result of two general types of oxygen starvation: “Hypoxia can be perfusion limited (acute hypoxia)…and can be diffusion limited (chronic hypoxia).” While this statement may serve as a working formulation for a first approach to the hypoxia problem, we would like to direct the readers' attention to two recent articles, which address the scope of this topic in a more comprehensive manner (1, 18). As a case in point, there are at least two models of diffusion-limited oxygen supply, depending on the microvascular geometry (Krogh model vs. Hill model). Referring to the seminal work of Thomlinson and Gray (12) in Section II.A, the authors outline that “lung carcinoma rods were surrounded by a necrotic core caused by a tissue oxygen gradient,” which may easily be mistaken as an example of a Krogh-type diffusion situation. Conversely, the model described by Thomlinson and Gray (12) actually describes necrotic areas in the center of the tumor cell rods, which are surrounded by blood vessels, a typical constellation corresponding to the Hill-type model.

According to Section II.A, “recurring tumors often exhibit a higher hypoxic fraction than primary tumors.” In fact, our group was the first to systematically measure and compare the oxygenation status of primary and recurrent tumors and, importantly, carried out these measurements in the same patients (2). This strategy enabled us to substantially support our hypothesis that hypoxia drives tumor progression and resistance to therapy since we unequivocally found a higher probability of more severely hypoxic primary cervix cancers to recur locally compared to less hypoxic tumors.

In Section II.B.4, the authors depict an inadequate “blood supply that further worsens local hypoxia, ensuring a vicious cycle.” Regarding the idea of a vicious cycle in hypoxia-associated pathophysiology, introduced into the oncologic field by our group in 2004, the readership may benefit from additional information regarding this complex concept given in the original articles (13, 19). In these sources, the role of the chaotic neovascularization on the efficacy of tumor blood flow and on the development of a hostile tumor microenvironment has been extensively discussed.

In Section III, the authors delineate that the direct methods for detecting tissue hypoxia are “providing oxygen concentration data.” This only holds true when using bare electrodes, which are not suitable for in vivo measurements in tissues due to protein deposits on the cathode. The statement is not correct, however, considering membrane-covered oxygen electrodes. It is the latter type of microelectrodes that has been used in tumors in situ exclusively (according to a protocol which has been implemented in the clinical and preclinical settings using standard procedures under well-defined boundary conditions by our institution). These membrane-covered microelectrodes rather measure oxygen partial pressures (pO₂ values, oxygen tensions). A conversion of pO₂ values measured in tissues to oxygen concentrations using Henry's law (pO₂=αO₂×cO₂, where pO₂=oxygen partial pressure, cO₂=oxygen concentration, αO₂=Bunsen's solubility coefficient) is not possible since Bunsen's solubility coefficient is not known for living tissues.

In Section III.A.1, the authors discuss several descriptive parameters of the pO₂ histogram. The use of hypoxic fractions, mean, and median pO₂ values have been suggested by Thews and Vaupel (11) for the clinical setting based on practical features of the Eppendorf histography system and hypoxic cutoff values for cellular and tissue functions as published in a comprehensive review later (3). In this section, Dr. Walsh et al. correctly argue that “the construction of three-dimensional (3D) oxygen maps is difficult.” Despite the fact that 3D oxygen tension maps have been published by Thews et al. (10) utilizing systematic pO₂ measurements within numerous electrode tracks and tissue planes, this approach is indeed unlikely to leave the field of experimental tumors in the near future.

In Section III.B.1–3, the authors acknowledge the fact that negative studies regarding the role of CA IX as a hypoxia marker exist (owing to its lack of correlation with both invasive pO₂ measurements and pimonidazole staining). However, the overall message these paragraphs convey seems to be that HIF-1α, CA IX, and GLUT-1 are established endogenous markers of tumor hypoxia. Unfortunately, several studies from our (6, 8, 9) and other laboratories (4, 5) have proven that this expectation will not be met. Although the aforementioned proteins often show an association with the local tissue oxygenation status, this association is not of a quantitative type. Possible explanations for these findings have been summarized in a review (7). In Section III.C.8, Dr. Walsh et al. describe the use of noninvasive positron emission tomography (PET) imaging for hypoxia detection. In general, we agree with their statements. However, there is a pivotal problem when using these techniques due to the fact that very steep pO₂ gradients exist (e.g., from 40 mmHg down to 0 mmHg within a distance in the order of 100 μm) causing hypoxic or even anoxic tumor microareas (“hypoxic niches”), which cannot be adequately sampled by current PET technologies because of spatial resolution limitations. Importantly, hypoxic microregions in tumors, not detectable in PET imaging, greatly contribute to tumor progression and acquired treatment resistance to anticancer therapies.

We agree with the authors that the assessment of (at least) pretherapeutic oxygenation status of individual tumors in the clinical setting would be desirable (i) to assign patients with hypoxic tumors to individualized treatment protocols in the setting of clinical trials, (ii) to develop and validate hypoxia modification therapies, and (iii) to intensify the post-therapeutic follow-up of patients with hypoxic tumors on a regular basis.

Finally, we congratulate Dr. Walsh et al. on an excellent and comprehensive review on the clinical importance of tumor hypoxia and unequivocally agree that the lack of hypoxia assessment in formerly conducted clinical trials “has made it difficult to extract the maximal benefit of hypoxia modification therapies for patients.”