The Practicalities of Angiogenic Doping

Because of the critical role of VEGF-A in the angiogenic response to endurance training, could VEGF-A be misused for angiogenic doping? Furthermore, could other interventions known to upregulate the VEGFA gene be exploited to achieve angiogenic doping indirectly?

The Current State of Pro-Angiogenic Therapies

The same technologies developed for gene therapy in ischemic diseases could be utilized for doping, in the hope of improving athletic performance. Currently, the use of VEGFs and other pro-angiogenic growth factors for gene doping is generally considered theoretical only. The first clinical trials targeting ischemic hearts were conducted more than 30 years ago by the research group of Jeffrey Isner [196–199]. The understanding of what does and does not work for pro-angiogenic therapy has increased over the years [200, 201]. Although no FDA-approved drug or procedure exists yet, the technology has been refined through successive clinical trials to the point that many experts are cautiously optimistic that the first breakthroughs may occur soon [202–204]. The key features that characterize most ongoing phase II and III trials are the use of secreted angiogenic factors (VEGF-A, VEGF-D, FGF, HGF), local highly efficient viral or mRNA delivery, and preclinical large-animal models (pigs) to optimize dose, safety, and pharmacology [201]. The technological barriers to achieving clinically meaningful delivery—such as the need for multiple targeted injections of high-titer viral vectors via catheter—currently limit the application of this technology to therapeutic use in patients. However, as treatment methods evolve and become more accessible, there is a potential risk that such technologies could eventually be exploited for performance enhancement.

Erythropoietin ahead of VEGFs in the mRNA Doping Landscape

Currently, commercially available gene doping products likely do not contain any VEGF-A or VEGF-D coding sequences, but have been shown to contain erythropoietin complementary DNA (cDNA). However, the amount of genetic material was deemed insufficient to produce an effect [205]. It is noteworthy that among the first commercially available off-the-shelf lipid nanoparticle (LNP)-encapsulated mRNA products were formulations designed to produce human erythropoietin upon injection [206, 207]. LNP-encapsulated mRNA is the technology that first debuted with the Moderna and BioNTech/Pfizer SARS-CoV-2 vaccines [208]. Depending on the injection site, different cell types would produce the erythropoietin (hepatocytes after intravenous injection, muscle cells after intramuscular injection, etc.). There are no published data on whether erythropoietin produced from synthetic mRNA in these cells can be distinguished from erythropoietin originating from mRNA transcribed from endogenous genes by its glycosylation patterns. In any case, the COVID-19 vaccines and mouse experiments with Epo-mRNA-containing LNPs demonstrate that this technology produces functional protein [209].

Doping and Anti-Ischemic Therapy Do Not Have the Same Goals

Do the modest results and technological barriers mean that angiogenic gene therapy would not work for doping? Not necessarily, because the treatment goals and the target population for gene therapy for ischemic diseases are quite different from those for angiogenic doping. Highly trained young athletes presumably have a different response to angiogenic stimuli than the predominantly elderly population with coronary artery disease. While small improvements in capillarization and oxygen extraction efficiency (such as has been shown by Mortensen et al. in untrained males [210]) might not be clinically meaningful in a patient population, such differences may be significant for athletes, where differences are measured in tenths of a second.

VEGF-A-Based Biological Drugs to Treat Ischemia

Neither the FDA nor the EMA have approved any VEGF-A therapy (protein, mRNA, or gene) for ischemic diseases. However, a gene therapy that delivers the VEGFA gene has been approved in Russia, under the name cambiogenplasmid (marketed as Neovasculgen®), for the treatment of peripheral arterial disease [211]. Its ‘naked plasmid’ technology is comparable to that used in phase II and III trials by the Isner group 30 years ago, none of which showed definitive clinical benefits, most likely due to very inefficient gene delivery [212].

Presumably similarly unsuitable for angiogenic doping purposes is telbermin (also known as sNN0029 in the context of amyotrophic lateral sclerosis treatment), an earlier attempt by Genentech to use VEGF-A protein as a pro-angiogenic drug to treat diabetic foot ulcers, which failed to meet the required efficacy in phase II. As a topical application, telbermin was designed to have a local effect only, relying on the wound to enter the body and unable to penetrate healthy skin [213, 214]. Previously, the same recombinant human VEGF-A had been used via intracoronary delivery in the VIVA trial for myocardial ischemia/coronary artery disease, with equally unimpressive results, likely due to poor uptake from the circulation, a short half-life, and the dose-limiting blood pressure-lowering effect of VEGF-A [215].

Merely changing to an intramyocardial delivery strategy did not improve the outcomes, with the REVASC (adenoviral VEGF-A121) and NORTHERN (intramyocardial plasmid VEGF-A165) trials similarly ineffective [216, 217]. Distinct from previous VEGF-A-based therapies is the currently still ongoing phase IIb trial to treat refractory angina with encoberminogene rezmadenovec (‘Adenovirus XC001’). This gene therapy produces three different VEGF-A isoforms (VEGF-A121, VEGF-A165, and VEGF-A189) from a single hybrid sequence, more closely mimicking the endogenous isoform distribution [218, 219].

Gene Therapy and Gene Editing

Gene therapy is the delivery of external genetic information into cells, causing them to produce the protein of interest. Gene delivery can be achieved in different ways. Currently, viruses are the vectors of choice because they have evolved to efficiently transfer genetic material into cells [217], and many studies on angiogenic growth factor gene delivery have used viral vectors [200, 202, 204, 220–222]. However, alternative non-viral vectors do exist, such as liposomes, which have been used to deliver mRNA in the Comirnaty and Spikevax Covid-19 vaccines [223]. Vectors can be administered directly (e.g., by injection) into target tissues, such as skeletal or cardiac muscle, in a process referred to as in vivo gene therapy [224]. Alternatively, target cells may be genetically modified outside the body (ex vivo) under controlled culture conditions before reintroduction into the patient [225, 226]. In addition to gene delivery, endogenous genes can be selectively altered through gene editing techniques. Since its introduction in 2012, the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein (Cas) system has rapidly replaced previous gene editing technologies in preclinical and clinical studies [227]. Although there is only one FDA/EMA-approved drug on the market [226], the theoretical possibilities of gene editing for both therapeutic and doping purposes are enormous, as reflected by approximately 250 ongoing clinical trials [228]. One major implementation hurdle remains the effective and specific delivery of CRISPR agents into target cells. Ex vivo editing of blood cells and liver targeting are low-hanging fruits expected to cover most of the CRISPR drugs anticipated to receive market authorization over the next few years [229]. However, possibly superior technologies based on CRISPR-related systems are being discovered and developed [230].

Which VEGF Would Work Best?

Among the angiogenic growth factors, VEGFs are considered the most specific, with VEGF-A and VEGF-D being the most promising targets for pro-angiogenic gene therapy [2]. While VEGF-A is the primary angiogenic factor during embryogenesis [42, 43], recent research suggests that other VEGFs may be superior during adulthood, particularly VEGF-D [79, 202, 204]. Specifically, the mature form of VEGF-D (also called ΔNΔC-VEGF-D) might be a superior option, because—unlike VEGF-A and VEGF-B [231]—it does not interact with VEGFR-1 and thus does not attract large amounts of inflammatory immune cells to the target tissue, and also is much more soluble since it does not—unlike the main isoform of VEGF-A—strongly interact with extracellular matrix and cell-surface associated proteoglycans [232]. Interestingly, active VEGF-D exists endogenously in two primary forms, one that is almost exclusively angiogenic [139] and another that supports both angiogenesis and lymphangiogenesis [78, 140], as reviewed by [55]. In addition, enhanced versions of VEGF-D with improved biological activity have been developed [233]. Lymphangiogenesis, which VEGF-C and VEGF-D can stimulate, appears beneficial in the experimental therapy of ischemic mouse hearts [234], and VEGF-C acts as a growth factor for the coronary vasculature [235]. Exercise increases the capillary filtrate, which needs to be drained, which is the primary task of the lymphatics [236, 237], but the effect of stimulating lymphangiogenesis on athletic performance is unknown. Apart from their angiogenic function, both VEGF-A and VEGF-C reportedly affect blood cell formation at the hematopoietic stem cell level [44, 238], and VEGF-A in the erythropoietic lineage [239]. Notably, based on mouse studies, VEGF-A has been proposed as a direct non-canonical, hypoxia-independent inducer of erythropoietin expression in perivascular stromal cells [240].

Would Hypoxia-Inducible Factor-1α be Superior to VEGFs in Boosting Tissue Oxygenation?

The use of any single angiogenic factor might not be sufficient to establish a physiological vascular network. Because the transcription factor HIF-1 acts as a master regulator of a large number of hypoxia-responsive genes, including VEGFA (see Sect. 2.1), it might provide superior vascularization results compared with the angiogenic factors themselves [241–243], also due to the cell type-specific responses [115, 244]. A study by Elson et al. [245] demonstrated that the expression of a stable HIF-1α mutant transgene in mice led to the formation of an organized, functional blood vessel network. HIF-1’s suitability is further supported by the effects on the vascular system demonstrated by HIF-stabilizing drugs. HIF-stabilizing drugs target the same genes as HIF-1α itself, including EPO and VEGFA. PHD inhibitors, such as roxadustat, which is approved for the treatment of renal anemia in numerous countries [246], target not only erythropoietin but also upregulate VEGF-A levels in animal studies [247]. However, the effects of PHD inhibitors on angiogenesis and vascular parameters, such as capillary density and branching patterns, have not been reported in any of the clinical trials. Roxadustat doping was first seen in 2015, before the drug had received market authorization [248]. The emergence of structurally diverse roxadustat analogs poses challenges for standard detection by liquid chromatography-tandem mass spectrometry (LC–MS/MS) [249]. A method to detect all HIF stabilizers in a single assay, based on their common function, has been developed but may not yet be sensitive enough [250]. Beyond LC–MS-based methods, alternative rapid screening approaches are emerging. For example, a recently described surface-enhanced Raman spectroscopy (SERS) platform enabled sensitive detection of PHD inhibitors in aqueous media and saliva at sub-µg/L levels [251].

Hypoxic Mimicry Targets Both Red Blood and Endothelial Cells

A chemically diverse range of chemical compounds (‘hypoxia mimetics’) can simulate low-oxygen conditions in the presence of normal oxygen levels and phenocopy the cellular and physiological responses to hypoxia. Such agents are widely used in research, as true hypoxia chambers for cell culture or animal husbandry are relatively expensive and cumbersome to use [252]. Because of the relative ease of access, hypoxia mimetics have already been experimented with for doping purposes as substitutes or enhancers of high-altitude training. In the following, we discuss carbon monoxide (CO), but it is only one of many compounds, including, among others, xenon and cobalt [253–255].

Unlike the unstable gasotransmitter nitric oxide (NO), which endothelial cells deploy as a signaling molecule in angiogenesis [256, 257], CO is stable and has been researched as a pharmacological agent [258–261]. Limited exposure to CO or CO-releasing molecules (CORMs) might be a means of increasing vascularization. In cultured vascular smooth muscle cells, which are a source of endogenous VEGF-A, a 1% CO atmosphere resulted in a very large 20-fold increase in VEGF-A generation [262]. While high concentrations of CO inhibit erythropoietin expression in cell culture [263], much lower concentrations can still inhibit prolyl hydroxylases and displace O2 from hemoglobin, resulting in hypoxia, which in turn upregulates erythropoietin and other hypoxia-responsive genes, such as VEGFA. Unsurprisingly, CO has emerged as a new doping agent [264]. As CO is highly lethal when overdosed [265], CORMs are considered safer than direct CO administration, and their action could be made more specific by coupling them to cell-specific antibodies to target their activity to particular organs or cell types, such as endothelial or smooth muscle cells. A safe CO delivery device (Covox DS) was developed by Ikaria, Inc. more than 20 years ago. Still, direct CO application was largely abandoned in favor of CORMs. In 2015, Ikaria was acquired by Mallinckrodt Pharmaceuticals, a company that later became infamous for its role in the US opioid crisis [266]. The use of CO is still somewhat common in exercise research as ‘CO rebreathing’ is an established research method for measuring the total hemoglobin mass in athletes [267]. CO—and, for that matter, all hypoxia mimetics when used systemically—not only target angiogenesis but also upregulate other hypoxia-responsive genes, most notably EPO. While most hypoxia mimetics have not been tested for effects on athletic performance, there is moderate evidence that some, including CO, may improve endurance performance [264].

No Angiogenesis Response to Moderate Hypoxia without Exercise

Two or eight weeks of 4100-m altitude-induced increase in HIF-1α expression were not sufficient to increase VEGF-A mRNA expression or capillary density. However, the subjects in this study were not athletes, but rather regular people who continued their usual level of physical activity at altitude [268]. At an altitude of 4100 m, the oxygen pressure is reduced from ~ 21 kPa (sea level) to about 12.8 kPa [269]. While this is a 40% reduction, it is not comparable to the decrease in oxygen pressure at the muscle cell level during exercise, which can drop from ~ 4 kPa at rest [270, 271] to approximately 0.3–0.5 kPa during heavy exercise [270, 272], with mitochondrial PO2 potentially falling below 0.13 kPa (1 mmHg) under maximal exercise conditions [273].

The results of Lundby et al. [268] indicate that, on average, the cardiovascular reserve of most people is sufficient not to require any neovascular response to moderately high altitudes. They also agree with our shared experience of air travel: cabin pressure, typically corresponding to an altitude of 2400 m, does not appreciably cause breathlessness in most people, even though the oxygen pressure is reduced by more than 25%.

In rat studies, moderately low oxygen levels (12%, equivalent to ~ 4400 m of altitude) did not induce neovascularization in largely inactive muscles, but did so in active muscles. Interestingly, regional heterogeneity in angiogenesis was not only observed between different muscles but also within a single muscle [274, 275], discussed in detail by [276]. Likely, angiogenesis does not result solely from increased transcription of VEGF-A mRNA, but translational control mechanisms might also be at work [98, 277], which would not be reflected in the mRNA levels. After the angiogenic adjustment of capillary density, which occurs relatively early in endurance training [167], the VEGF-A concentrations required to maintain a vascular network are likely very similar, largely independent of its density.

Pseudoanemia

The finding that some endurance athletes have relatively low hematocrit and hemoglobin levels is sometimes referred to as “sports anemia” or “pseudoanemia.” Despite relatively low hemoglobin and RBC concentrations, these athletes can have a large total hemoglobin mass due to their large total blood volume. While this phenomenon is not observed in all athletes, it is widespread because blood volume expansion is an early and consistent response to endurance training [278–281].

To what degree and by which mechanisms pseudoanemia contributes to improved performance has been controversial. Enhanced cardiac performance [282], reduced blood viscosity [283] and cardiovascular strain [284], as well as increased thermoregulatory capacity (more blood flow to the skin to cool down) [285, 286] have been proposed to contribute to the effect. Specific to runners is the controversial ‘footstrike hypothesis’, according to which older, less functional RBCs are mechanically destroyed during footstrike, thereby rejuvenating the RBC population [287–289]. Interestingly, the possible beneficial role of blood volume expansion as a compensatory mechanism has been re-examined, even in pathological contexts such as heart failure [290].

Beyond its relevance for masking erythropoietin doping, pseudoanemia could have significant implications for detecting angiogenic doping via the ABP. When new vessels form or existing vessels dilate without concurrent RBC stimulation, compensatory plasma volume expansion would decrease hemoglobin and RBC concentrations, creating a detectable hematological signature analogous to post-donation hemodilution.

Risks and Problems Associated with the Execution of Angiogenic Doping

If pro-angiogenic responses are achieved through gene delivery or gene editing, all commonly known risks associated with such therapies become relevant [225, 291, 292]. While today’s regulatory requirements ensure relative safety and the highest possible quality in vector production [293], this is not the case when such tools are produced in rogue labs or administered outside a regulated healthcare system. Off-target effects can result from viral integration [8] or spurious annealing of the CRISPR guide RNA [292]. Repeated injections may induce immune reactions [225], and local overdosing may occur [294]. In addition to these generic side effects, specific side effects are associated with the overexpression of any particular gene. For angiogenic growth factors, concerns have included the vascularization of dormant tumors and other forms of aberrant angiogenesis (VEGF-A, VEGF-D) [83, 84, 154], heightened immune responses (VEGF-C) [295, 296], and inflammation (VEGF-A, VEGF-B) [231, 297]. Atherosclerosis, neovascular eye disorders, and many other diseases have been associated with the overexpression of VEGFs [156, 298]. It is unknown currently whether there would be any specific risks associated with a doping-induced denser vascular network, similar to the particular risk of erythropoietin-induced hyperviscosity syndrome (‘blood sludging’) leading to blood clots and possibly even sudden cardiac death.