Although hyperbaric oxygen therapy (HBO) has been in the armamentarium of diabetic foot management for almost 50 years , the acceptance of HBO among clinicians treating patients with diabetic foot ulcers has been poor. Inevitably, HBO has been used by charlatans unscrupulously promoting cure of almost any disease, and evidence supporting the usefulness of HBO in diabetic foot ulcer healing has been limited, although not inferior to many other more commonly used treatment modalities in the field of the diabetic foot [2, 3]. In recent years, two double-blind randomized controlled trials (RCTs) have shown beneficial effect of HBO in terms of ulcer healing and health-related quality of life [4-6].
In this review, the rational and clinical setting of HBO and the present clinical evidence for the use of HBO in the armamentarium of the diabetic foot are discussed.
From a pharmacological point of view, HBO could be described as a short-term, high-dose oxygen inhalation and diffusion therapy, delivered systemically through airways and blood, achieved by having the patient breathing concentrated oxygen at a pressure higher than 1 absolute atmosphere (ATA) . In clinical practice, hyperbaric chambers are used to achieve this. Breathing high-dose concentrations of oxygen at surface levels and topical exposure of limbs to high-dose concentrations of oxygen are not included in the definition of HBO.
Mechanisms of action
Hyperbaric oxygen therapy exerts its therapeutic effect by four mechanisms: mechanical effects, bacteriostatic effects, hyperoxygenation for treatment of monoxide and cyanide poisoning, and finally, treatment aiming for restoration from hypoxia .
The principles for the use of HBO are based on application of the basic physical laws in effort to correct abnormal tissue oxygen tension. Boyle's law states that at constant temperature, the volume of a gas is inversely proportional to its pressure; Dalton's law states that the total pressure exerted by a gaseous mixture equalizes the sum of the partial pressures of each individual gas in the mixture; and finally, Henry's law states that the concentration of gas in a solution is determined by its partial pressure and by its solubility coefficient.
Under normal perfusion conditions, resting tissues extract about 50 mL of oxygen per litre of blood, most of which is delivered by haemoglobin. During air breathing at normobaric pressure, haemoglobin is saturated to about 97% when leaving the pulmonary circulation, and only a small fraction of oxygen is dissolved in the blood, about 3 mL of oxygen per litre of blood. If 100% oxygen is administrated, the volume of dissolved oxygen increases to 15 mL per litre of blood, and during HBO at 2.5 ATA, almost 60 mL of oxygen is dissolved in each litre of blood .
After transportation to the capillary bed, oxygen is dependent on diffusion to reach cells and intracellular spaces. The rate of oxygen delivery is inversely proportional to the square distance and directly proportional to pO2 at the initial point at the capillary .
The intercapillary distances vary between different tissues. A major denominator of capillary density is metabolic rate; that is, distances between capillaries are short in muscle and other highly oxygen-consuming tissues but considerably longer in slow-healing tissues with lower metabolic rates, such as tendon, fascia, and subcutis. In diabetic microvascular disease, capillary function is declined and distances between capillaries are increased. To avoid hypoxia in the presence of microvascular disease, oxygen needs to diffuse longer distances, thus requiring higher pO2 levels at the edge of the capillaries. The increment in pO2 at therapeutic hyperbaric conditions creates a potential threefold augmentation in diffusion distance .
Rational of HBO in ulcer healing
Hyperbaric oxygen therapy has been shown to improve leukocyte function, stimulate vascularization, and enhance formation of granulation tissue [12-14].
In wounds, the single most bactericidal mechanism is oxidant production, and pO2 is one of the most important factors for bactericidal production in human leucocytes . HBO facilitates the oxygen-dependent peroxidase system in leucocytes, increases generation of oxygen free radicals and thereby enhances oxidation of proteins and membrane lipids, and inhibits bacterial metabolic function [16, 17]. In ischemic tissue, restoration of tissue pO2 re-establishes the phagocytic function of macrophages [18-20]. HBO improves oxygen-dependent transport of certain antibiotics across bacterial cell walls [21, 22]. Also, environment becomes less suitable for anaerobic bacteria as pO2 increases in the tissue.
Neovascularization occurs through two mechanisms, angiogenesis and vasculogenesis. The former is stimulated by regional factors and the latter by recruitment and differentiation of circulating stem/progenitor cells (SPCs) [23, 24]. HBO enhances these processes, that is, by increasing production of growth factors such as vascular endothelial growth factor, which is the most specific growth factor for neovascularization [25, 26]. Oxidative stress at sites of neovascularization stimulates growth factor synthesis by augmenting synthesis and stabilization of hypoxia-inducible factor 1, a process further enhanced by HBO [27-30]. HBO also enhances extracellular matrix formation, an O2-dependent process closely linked to neovascularization [31, 32].
Nitric oxide synthase 3 activity is required for SPC mobilization from the bone marrow. SPC mobilization is impaired in patients with diabetes, probably because of reduced nitric oxide synthase activity caused by hyperglycaemia and insulin resistance [33, 34]. HBO mobilizes SPC in people with and without diabetes by stimulating NO synthesis in bone marrow [28, 35]. HBO reduces tissue oedema by mechanisms of vasoconstriction in nonischemic tissue .
High lactate concentrations in wounds stimulate procollagen synthesis . The enzymatic steps in which procollagen is converted to collagen and cross-linked to collagen matrix are oxygen dependent . In animal experiments, collagen deposition in hyperoxic environment has been shown to be three times that in a hypoxic environment. Correction of both vasoconstriction and hypoxemia might increase collagen deposition tenfold . Furthermore, fibroblast replication is most optimal at tissue pO2 levels of 40–60 mmHg .
Clinical setting of HBO
There are two different kinds of hyperbaric chambers, monoplace and multiplace chambers. A monoplace hyperbaric chamber is generally made of acrylic material to permit direct patient observation. The cylinder is usually pressurized entirely with oxygen. Multiplace chambers are typically steel constructions in which two or more patients are pressurized. For safety reasons (fire hazards), the ambient gas is pressurized with air and patients breathe oxygen via hoods or masks.
Robust evidence is lacking for selection of treatment regimen leading to optimal therapeutic benefit (i.e. hyperbaric pressure level, duration of treatment sessions, number of HBO sessions, and not least timing of HBO). Patients with diabetic foot ulcers are usually treated once daily at pressures of 2.0–2.5 ATA . During a treatment session, patients usually breathe oxygen for 80–90 min. Another 5 to 10 min per session is generally required for compression and decompression. To minimize the risk of oxygen toxicity – a rare but severe complication of HBO – we may separate periods of breathing 100% oxygen by one or two 5-min-long intervals of air breathing. A typical treatment protocol consists of 30–40 treatment sessions .
The only absolute contraindications for HBO are untreated pneumothorax, ongoing treatment with some chemotherapeutic agents, and any history of treatment with bleomycin. Relative contraindications are sinusitis, severe chronic obstructive pulmonary disease, history of pneumothorax or thoracic surgery, uncontrolled high fever, claustrophobia, upper respiratory infection, and inability to equalize pressure in the middle ear. Many of these contraindications are related to known complications of HBO, such as barotrauma, which can be exacerbated by emphysema or inability to equalize middle ear pressure, and seizures for which an uncontrolled fever can be a predisposing factor.
Evidence of clinical usefulness
The first study presenting data on diabetic foot ulcer outcome after HBO was published in 1979 by Hart et al., the first controlled study in 1987 by Baroni et al., and the first RCT 5 years later by Doctor et al. [48-50]. Another five RCTs evaluating the effect of HBO in patients with diabetic foot ulcers have been published as well as several case series and nonrandomized controlled studies (Table 1).
|Reference and year||Wound description||No. of patients||Outcome and comments|
|Hart and Strauss, 1979 ||Chronic nonhealing diabetic foot ulcers||11||Healing rate, 10/11 (91%)|
|Matos, 1983 ||Nonhealing diabetic foot ulcers||70||Healing or significant improvement was seen in 60% of all patients|
|Ischemia, infection, or neuropathy present|
|Baroni et al., 1986 ||Wagner grade 3 and 4||I: 18||Healing: I: 16/18; C: 1/16|
|Matched control group||C: 16||Amputation: I: 2/18; C: 4/16|
|Surgeons blinded||Amputation rates at the centre, 40%|
|Perrins and Barr, 1986 ||Unknown||26||Healing rate, 67%|
|Amputation was avoided in 18%|
|Davies, 1987 ||Wagner grade 3 and 4||168||Healing rate, 70%|
|Daily HBO for 30–60 days|
|Wattel et al., 1990 ||Ulcers, 11 diabetic and 9 arteriosclerotic||20||Healing was seen in 15 patients|
|TcPO2 was a predictive factor for ulcer healing|
|HBO, 2.5 ATA|
|Two sessions of 90 min per day|
|Median, 46 (15–108) sessions|
|Oriani et al., 1990 ||Controls were selected for, but refused, HBO||I: 62||Healing: I: 66%; C: 33%|
|Group extension of Baroni's study?||C: 18|
|Cianci et al., 1991||Wagner grade 4||41||Limbs were salvaged in 78%|
|Limb-threatening infection in 97%|
|Revascularization in 55%|
|Doctor et al., 1992 ||Chronic ulcers||RCT||Amputation: I: 2/15; C: 7/15, p < 0.05|
|Randomization after necessary debridement and 3 days of antibiotic treatment in hospital.||I: 15|
|Four HBO sessions, 45 min, 3.0 ATA|
|Weisz et al., 1993 ||Ulcer duration >3 months||14||Healing was seen in 11 patients|
|All palpable foot pulses|
|HBO, 2.5 ATA|
|Median, 56 ± 10 sessions|
|Ciaravino et al., 1996 ||Nonhealing wounds||54||Some improvement, 11%|
|Diabetes, 17/54 patients||No improvement, 80%|
|Mean, 30 HBO sessions||Inconclusive, 9%|
|Complications in 63% (barotraumas, 43%)|
|Faglia et al., 1996 ||Wagner grade 2–4||RCT||Amputation: I: 8.6%; C: 33.3%, p < 0.02|
|Duration >3 months||I: 36|
|HBO, 90 min, 2.2–2.5 ATA||C: 34|
|Lee et al., 1997 ||Infected foot ulcers||31||Major amputation, 6|
|HBO, 35 ± 22 sessions||Healing, 25|
|Zambroni et al., 1997 ||HBO, 2.0 ATA, 30 sessions||I: 5||Healing at 4–6 months of follow-up:|
|C: 5||I: 80%; C: 20%|
|Kalani et al., 2002 ||Basal TcPO2 <40 mmHg and >100 mmHg breathing O2||I: 17||Three-year follow-up:|
|C: 21||Healing: I: 76%; C: 48%|
|Amputation: I: 12%; C: 33%|
|HBO, 2.5 ATA, 40–60 sessions|
|Kessler et al., 2003 ||Wagner grade 1–3||RCT||Ulcer area reduction:|
|Duration >3 months||I: 14||Week 2: I: 42%; C: 22%, p < 0.04|
|Neuropathy present||C: 13||Week 4: I: 48%; C: 42%, NS|
|2 sessions/day, 10 days|
|90 min, 2.5 ATA|
|Abidia et al., 2003 ||Ulcer duration ≥6 weeks||RCT||One-year follow-up:|
|Diameter, 1–10 cm2.||I: 9||Ulcer healing: I: 63%; C: 0/8, p < 0.03|
|ABI <0.8 or TBI <0.7||C: 9||Amputation: I: 13%; C: 13%|
|HbA1c <8.5%||Drop out: I: 1; C:1|
|HBO, 30 sessions, 90 min, 2.4 ATA|
|Duzgun et al., 2008 ||Ulcer duration ≥4 weeks||RCT||Ulcer healing without debridement in the operating room: I: 33; C: 0|
|HBO, 30–45 sessions, 90 min, 2.5 ATA||I: 50|
|Löndahl et al., 2010 ||Ulcer duration >3 months||RCT||One-year follow-up|
|No need for or possibility of vascular surgery||I: 49||ITT analysis|
|HBO, 40 sessions, 90 min, 2.5 ATA||C: 45||Ulcer healing: I: 52%; C: 29%, p = 0.03|
|Amputation: I: 7%; C: 3%|
|Per protocol analysis|
|Ulcer healing: I: 61%; C: 27%, p < 0.01|
|Amputation: I: 3%; C: 3%|
Although case series and nonrandomized studies constitute a limited source of evidence, the consistency in outcome between these studies is notable, although publication or patient selection bias or other confounding factors could be involved.
Previous reviews of RCTs evaluating the effect of HBO have identified several limitations in study design and methodology, including lack of blinding, limited information on randomization, study procedures, wound classification, and specified use of sham therapy [3, 51-55]. Only two studies score all 5 points on the Jaded scale (Table 2).
|Doctor et al., 1992 ||Faglia et al., 1996 ||Kessler et al., 2003 ||Abidia et al., 2003 ||Duzgun et al., 2008 ||Löndahl et al., 2010 |
|Randomization||Not described||Not described||Not described||Described||Described||Described|
|Conclusion||The number of patients in each group is not sure||Uncertain because of the high number of vascular interventions||Yes, but one patient was excluded because of barotraumatic otitis||Yes, one patient in each group excluded before treatment||Yes||Yes, intention-to-treat and per protocol|
|Blinding||No||Only surgeons||Blinded observer||Double blinded||No||Double blinded|
|Concomitant treatment||Unknown||Interventions post-HBO may interfere with outcome||Similar||Similar||Unknown||Similar|
|Baseline characteristics||Similar||Similar||Similar||Similar||Not similar||Similar|
|End point||Described, amputation||Described, amputation||Ulcer area reduction||Described,||Healing without surgical intervention||Described,|
|1. ulcer healing||1. ulcer healing|
|2. amputation||2. amputation|
|Time frame||unknown||until healing or amputation||4 weeks||1 year||until healing||1 year|
|Power||No analysis||No analysis||No analysis||No analysis||No analysis||Estimation|
|n = 30||n = 70||n = 28||n = 18||n = 100||n = 94|
Altogether, in published controlled trials with follow-up times between 3 months and 3 years, ulcer healing has been evaluated in 108 patients in double-blind RCTs and in 208 patients in nonrandomized controlled studies. Healing rates are significantly higher in patients treated with adjunctive HBO as compared with placebo or multidisciplinary wound care alone in RCTs (54% versus 24%) as wells as in nonrandomized controlled studies (77% versus 25%) [4, 5, 49, 59-61]. Major amputation rates are reported in four of these RCTs and three of these controlled but nonrandomized trials [4, 5, 49, 50, 56, 61]. Major amputations has been performed in 5% (n = 61) of HBO-treated patients and 24% (n = 57) of non-HBO-treated patients in the non-RCTs compared with 8% (n = 107) and 19% (n = 104) in the RCTs [4, 5, 49, 50, 56, 61]. However, amputation rates in the control groups of these studies were considerably higher in the earlier nonblinded RCTs (36% versus 4%) [4, 5, 50, 56]. This might mirror a change in indications for major amputation, because nonhealing of a chronic ulcer could be an indication of major amputation in the setting of the earlier but not in the later studies. The study by Duzgun et al.  is not included in this analysis because major and minor amputations were differently defined.
Several studies, including the two double-blind RCTs, show improved health-related quality of life in patients receiving HBO [4, 5, 62].
Although HBO has several potential side effects, compared with many other medical therapies, it may be considered as a rather safe treatment modality. Among the most commonly reported complications are barotraumas. Barotrauma may occur at any tissue and gas interface within the body. In clinical practice, middle ear barotraumas dominate, but the most important organ that may be affected is the lung. In a case series of 11 376 treatment sessions, 17% of all patients reported ear pain or discomfort during compression . However, persistent injuries visible in ear microscopy are less common, with reported incidences between 0.5% and 3.8% [5, 40, 41]. However, because barotrauma might lead to persistent hear loss or vertigo, tympanostomy with tube placement ought to be considered when patients have problems to equalize pressures.
During decompression, the intrapulmonary gas volume increases, and if this additional gas volume cannot be breathed out, lungs may tear from an overpressure, causing pneumothorax, emphysema, or in worst-case scenario, an air embolus. Pulmonary barotraumas are rare, with an incidence of 1 in 50 000 to 60 000 treatments [42, 43].
Reversible myopia, due to oxygen toxicity of the lens, is a common side effect affecting up to every fifth patient . Cataract is not a clinical problem during normal treatment series but seems to be persistent and highly frequent after prolonged treatment series (>150 treatment sessions) . Such exposures are therefore no longer recommended.
Patients with diabetes mellitus, especially those on insulin therapy, are at increased risk of hypoglycaemia, usually occurring within 2 to 6 h of the HBO session [5, 46].
Oxygen seizure is a rare and self-limiting complication without any long-term implications .
Prolonged exposures for oxygen may cause pneumonitis and alveolitis, but these consequences of pulmonary oxygen toxicity are not a problem in clinical use of HBO.
The most common fatal complication is associated with fire in the chamber. At least 88 human fatalities in 36 separate hyperbaric chamber fires have been reported . Increased fire risk applies especially to chambers pressurized with oxygen.
Cost-effectiveness is important in treatment selection. Only a few health economic analyses evaluating the cost-effectiveness of HBO in diabetic foot ulcer treatment have been published, and they are limited by deficient primary clinical data [4, 63-65]. Still, these studies suggest a potential cost-effectiveness of HBO, for example, a crude analysis of the double-blind RCT by Abidia et al.  – only taking HBO and dressing costs into account – suggests a saving of £2960 per patient during the first year of follow-up.
The cost of a full-course HBO treatment for diabetic foot ulcer varies from one place to another, and several factors including set-up costs, ongoing costs, reimbursement systems, and numbers of patients treated per centre have impact on the cost. Charges between $200 and $1250 per treatment session have been reported from reimbursed health-care units [4, 63, 65].