Theoretically, any portion of a limb or tissue where an arterial occlusion can be applied safely can be a measurement site. Typical measurement sites are: the forearm (brachioradialis or flexor digitorum profundus), calf (gastrocnemius or tibilias), hand (thenar eminence) and foot (Table 1). Consideration of the research question and participant population is important prior to muscle group selection. The thenar eminence has been recommended due to low levels of overlying subcutaneous adipose tissue thickness (ATT), minimal oedema, and good reproducibility in healthy subjects (Mayeur et al., 2011; Gómez et al., 2009). If an exercise intervention is being tested, a muscle group which is suitable for training should be chosen. Similarly, if the participant population is likely to have excess adiposity, a site with less ATT is preferable.

Different anatomical sites may differ in terms of metabolic and vascular properties, muscle thickness, fibre type proportion and overlying ATT. These differences make comparisons between studies challenging. PORH has been compared across different measurement sites in the same individuals by Gomez et al. (Gómez et al., 2009). They found that baseline oxy [heme] and the magnitude of oxy [heme] desaturation and oxy [heme] re-saturation were lower in the forearm compared to the hand; however, we do not know the extent to which these reflect differences in microvascular behaviour or differences in overlying tissue properties that impact NIRS measurements. Future work on these questions would be useful. Use of multi-channel devices at different points along the same muscle during PORH could permit intra-muscle similarities/differences to be examined. Consideration of distance between measurement site and the cuff occlusion site may also be important, but we are unaware of any published work that has explored this.

Participants who undertake PORH measurements often find the occlusions uncomfortable, and at times, painful. Lacroix et al. reported that the forearm occlusion was less uncomfortable than lower limb occlusion (Lacroix et al., 2012) but there has been limited work exploring the effect of discomfort on results; if investigators frequently find it necessary to exclude participants who are unable to complete the pre-specified occlusion duration, this may introduce selection bias into presented results. Future work exploring participant discomfort during arterial occlusions, perhaps via incorporation of pain scale questionnaires, could provide valuable practical insights.

If ATT is greater than half the distance between the NIRS source and detector, little or no signal from skeletal muscle will be detected (Van Beekvelt M. et al., 2001). Most NIRS devices have fixed distances of 2–4 cm, but some have multiple light sources or detectors fixed in series (multi-distance devices). Using longer source-detector distances allows greater sampling depth but at the cost of reduced signal.

Despite being a well-recognised limitation, ATT is not routinely measured or corrected for in studies that use NIRS. Measurement of subcutaneous ATT overlying the muscle of interest can be done using callipers or an appropriate ultrasound device. Out of the studies in Table 1, three included a correction method for ATT, one adjusted source-detector distance to at least twice that of an individual’s ATT (41), and two used statistical adjustment to account for ATT (36, 37). In a study where total [heme] (frequency domain NIRS) and ATT were measured across different muscles, results suggested ATT correction equations for oxy [heme] and deoxy [heme] may be specific to the muscle of interest (Craig et al., 2017). In work where the reperfusion response has been modelled using sigmoidal or exponential functions, Bopp et al. found that the calculated time constants representing the reperfusion curve (τ1) or the return to baseline curve (τ2) were far less affected by ATT than the parameters reflecting the magnitude of change (Bopp et al., 2014; Bopp et al., 2011). This suggests that modelling the kinetics of the PORH may be a useful way to circumvent the impact of ATT on the analysis endpoints. Further studies examining the effect of ATT on quantification of PORH would be useful.

The magnitude of the ischemic stimulus is an important determinant of PORH. The most common occlusion durations reported are three or 5 min. A limitation of using pre-set occlusion durations is that rates of deoxygenation may vary considerably across individuals. In a recent model, the StO₂ deficit was directly associated with the reoxygenation rate; as oxygen deficit increased, reoxygenation rate increased (Anderson and Park, 2023). McLay et al. (McLay et al., 2016a) found that in all of their participants, a desaturation plateau was reached before 5-min of occlusion, but 3-min occlusions in other studies have resulted in varying desaturation levels (Mayeur et al., 2011). A major advantage of NIRS, compared to other techniques, is that changes in tissue saturation can be recorded during the occlusion and a rate of desaturation (oxygen consumption rate) can be calculated. This allows users to standardize the ischemic stimulus or account for it in a post-processing step. Rosenberry et al. (Rosenberry et al., 2018) highlighted the potential importance of differences in desaturation. When they adjusted the occlusion duration in young and old individuals to standardize the level of desaturation, they found that age-related differences in PORH were attenuated. Some studies present the amplitude of oxy [heme] change as a percentage of the occlusion stimulus to circumvent this problem (Jarm et al., 2003; Kragelj et al., 2001; McGranahan et al., 2023). Adjusting occlusion duration so that a pre-specified reduction in oxy [heme] or StO₂ is achieved is an alternative and several studies have employed this approach (Monthé-Sagan et al., 2016; Mayeur et al., 2011; Zamparini et al., 2015; Fellahi et al., 2014). Taking baseline StO₂ value into account may also be beneficial, since this can differ between participants due to varying basal metabolic rates and vascular architecture at the measurement site.

Broadly speaking, a NIRS device works by emitting a known intensity of NIR-light (650–950 nm) at two or more wavelengths into tissue, measuring the intensity of returning light at each wavelength. Based on differential absorption patterns of each wavelength by oxy [heme] and deoxy [heme], the device estimates the concentration (or concentration change) of each chromophore (Jones et al., 2022). This simple technology has been developed into several different device types: continuous wave (CW), frequency-domain (FD) and time domain (TD) devices, reviewed elsewhere (Barstow, 2019b). Thus far, only FD and CW devices appear to have been employed for PORH measurements (Table 1). CW devices are simple and cost-effective but are limited in only providing information about relative chromophore concentrations. During PORH, as we are interested in the response to ischemia, this is not a major limitation. FD devices can account for light scattering and therefore measure absolute concentrations. Direct comparisons between different types of NIRS devices during PORH would be valuable.

Multi-distance CW-NIRS devices can use spatially resolved spectroscopy (SRS) to calculate a StO₂ (Artinis Medical Systems TN, 2022). SRS addresses some of the problems relating to variations in ATT, and StO₂ is frequently used for PORH. There are some variations in the design of devices applying SRS. Some devices use multiple light sources and a single detector while others use a single light source with multiple detectors to quantify linear change of light attenuation against distance (Suzuki et al., 1999). We have not found studies comparing different light source-detector arrangements for PORH. A source of confusion is that the term ‘StO₂’ is used synonymously with ‘TSI’ (tissue saturation index), ‘TOI’ (tissue oxygen index) and ‘SmO₂’ (muscle tissue oxygen saturation) in the literature. Also, some authors calculate StO₂ as oxy-[heme]/(oxy [heme] +deoxy [heme])*100% in the absence of SRS. Consensus on use of these terms has previously been called for and would be beneficial.

NIRS devices typically have high temporal resolution (>1 Hz), although sampling rates as low as 0.25 Hz have been applied successfully (Mayeur et al., 2011; Zamparini et al., 2015; Cheatle et al., 1991; Fellahi et al., 2014). In contrast, VOP is limited to one measurement every 5–10 s. Higher sampling frequency is beneficial for capturing changes that occur immediately on cuff release and allows identification of pulsatile changes in the oxy-Hb signal at rest. Reciprocal changes in oxy [heme] and deoxy [heme] signals and absence of pulsatility are useful criteria to indicate successful arterial occlusion.

The absorption spectra of oxy- and deoxy-myoglobin (Mb) within myocytes and circulating oxy- and deoxy [heme], are very similar. It has been estimated that, in contracting muscles, the Mb contribution to the signal ranges from 20% to 80% (Lai et al., 2009; Davis and Barstow, 2013). The possibility that the oxy [heme] contribution is not constant, changing with varying degrees of tissue hypoxia, has implications for PORH. This was recently explored through combining an experimental canine model with a computational model of muscle O₂ transport and utilization (Koirala et al., 2023). This analysis indicated that Mb contributed ∼20–30% of NIRS signals at rest, but when O₂ delivery was reduced, the proportion of deoxy-[heme] signal due to Mb increased to 50% and the contribution to the oxy [heme] signal increased from 0% to 30%. This suggests analysis endpoints derived using only the oxy [heme] NIRS signal would be influenced by changing Mb contributions throughout desaturation or resaturation. Considering changes in the total [heme] signal or a StO₂ signal would address this problem to some extent as it is unlikely that the total concentration of myocyte Mb changes during the ischemia-reperfusion manoeuvre.

Another consideration is that type I (slow twitch) muscle fibres typically contain higher concentrations of Mb than type II (fast twitch) muscle fibres (Bekedam et al., 2009). Variation exists in the typical fibre-type proportion between muscles; for example, the soleus muscle is known to be comprised of mostly type I fibres while triceps contain mostly type II fibres (Talbot and Maves, 2016). This could be important when comparisons are made across different muscle groups. However, it is also known that muscle fibre type can be altered with training, which may pose difficulties in comparing untrained with trained populations or pre-versus post-exercise intervention (Talbot and Maves, 2016). Recent work in elite cyclists suggests that they may overall have reduced Mb concentration compared to active controls (Jacobs et al., 2023). Furthermore, it has been shown that Mb concentration is greater in older compared with young adults (Möller and Sylven, 1981); this has implications for comparisons of PORH by age.

Several studies have reported the reproducibility of NIRS PORH(13, 32, 34, 41, 44, 56). Test-re-test reproducibility for temporal, absolute range and reperfusion recovery slope parameters ranges from adequate to excellent (intra-class correlation coefficients (ICC) between 0.5 and 0.9) in the leg (McLay et al., 2016b; Willingham et al., 2016; McGranahan et al., 2023) and adequate to good in the arm (ICC 0.6–0.84) (Lacroix et al., 2012). Time to peak hyperaemia was less reproducible than time to 95% peak, since the duration of the last 5% of reperfusion can vary substantially (Willingham et al., 2016). Mclay et al. directly compared repeatability of PORH by NIRS and FMD in the leg and found better intra- and inter-day reproducibility for the NIRS measurements (McLay et al., 2016b). These studies used 5-min occlusion durations and did not adjust for oxygen deficits (desaturation) during the occlusion. Mayeur et al. (Mayeur et al., 2011) reported better reproducibility when using a 40% desaturation target versus a set 3-min duration in the forearm (coefficient of variation 5.9% versus 10.5%). Further studies exploring reproducibility of NIRS-measured PORH following different occlusion durations will be valuable in establishing the precision of the technique.

As NIRS technology develops, it will be important to assess repeatability of newer devices for assessment of PORH. For example, use of StO₂% measured by SRS in place of raw oxy [heme] and deoxy [heme] data for PORH is becoming more widespread and it will be important to establish the reproducibility of StO₂% in quantifying PORH. Assessing the precision of more advanced modelling techniques for quantifying PORH, such as methods described by Bopp et al. (Bopp et al., 2014), will also be important.