Key result
Normalization of force to predicted oxygenated area yielded a lower coefficient of variation than normalization to total cross-sectional area, but did not improve upon raw data.
In isolated heart muscle preparations, normalizing force to cross-sectional area increases data scattering, and while normalizing to oxygenated area is better, it does not improve upon using raw force data.
Answers are grounded in this paper and cite their sources.
Normalization to total cross-sectional area should be avoided in isolated muscle studies; leaves open optimal force reporting methods for future research.
Based on the relationship between radius of the muscle tissue (R) and the fraction of whole cross-section of muscle supplied with oxygen (θ) defined by Han et al,1 the cross-section of muscle supplied with oxygen was calculated as: Aoxy = θ*π*R2. In the past, normalization of force to cross-sectional area gained much popularity but has severe limitations. First report about normalization to cross-sectional area originates from Kelly.2 They used a larger number (n = 70) of rat left ventricle papillary muscles with a radius between 0.25 and 1.2 mm. Muscle thickness was measured under pre-tension. Raw data for force were not presented, but force normalized to cross-sectional area correlated negatively to cross-sectional area. Koch-Weser3 studied the relationship between muscle cross-sectional area and force in very detail. They used rather thin papillary muscles from cat right ventricles. Muscle cross-sectional area was between 0.2 and 1.5 mm2 (= radius between 0.12 and 0.35 mm). Force normalized to cross-sectional area was constant up to an area of 1 mm2 (radius of 0.55 mm) when paced at slow rate (18.8 bpm) but declined already above an area >0.6 mm2 (radius 0.44 mm) when paced 10 times faster. From this finding, the authors concluded that the critical thickness depends on the actual work of the muscle. It remains unclear from that study what the critical radius at 1 Hz could be. However, on the basis of interpolation, we would expect a value between 0.44 and 0.55 mm. Findings of Koch-Weser were replicated in cat right ventricular papillary muscles by Fisher et al.4 They found a constant relationship between cross-sectional area and force (n = 4). This finding is at contrast to several other publications where researchers failed to detect any stable plateau for normalized force values even at very small radius. Bing et al5 had used rat left ventricular trabeculae (smallest radius 0.4 mm) and rat left ventricular papillary muscles (smallest radius 0.35 mm). Only Raman6 could demonstrate a stable plateau for force normalized to cross-sectional area in a population of very thin free running right ventricular trabeculae prepared from mouse heart. However, it should be noted that the critical radius was found to be 0.06 mm, much lower than reported in the above-mentioned studies.3, 5 Trying to overcome the limitation of normalization to total muscle area, we retrospectively analysed whether normalization of force of contraction to oxygenated area rather than normalization to total area reduces data scattering in isolated heart muscle preparations. We therefore analysed 856 human right atrial tissues from 124 patients paced at 1 Hz and 37°C. Isometric force of contraction was normalized to cross-sectional area or the area of muscles oxygenated as predicted from the computer model by Han et al. Cross-sectional area as well as force of contraction showed huge variability (Figure 1A). To investigate whether variability in force is related to differences in muscle thickness, we plotted individual force data against respective muscle thickness, expressed as cross-sectional area (Figure 1A). A simple linear regression curve was fitted to the data points. There was a weak positive correlation between cross-sectional area and basal force. However, the slope was rather low and the coefficient of determination R2 accounted to only 0.01. Both results argue against a simple relationship between muscle cross-sectional area and force. Even if the relationship between force and cross-sectional area is not perfect, normalization could reduce data scatter. Therefore, we have calculated coefficients of variation for raw data and data normalized for total cross-sectional area. The coefficients accounted to 0.818 and 1.079. Variability was statistically significant higher (P < 0.001) when data were normalized to cross-sectional area than raw data (two-sided P-values for differences between coefficients of variation where estimated using bootstrapping). Next, we wanted to evaluate whether our findings may represent a peculiarity of human atrial trabeculae. For that purpose, we used data from guinea-pig right ventricular papillary muscles, published by one of us.7 The muscles were conical in shape like human atrial trabeculae, but slightly thicker (radius 0.76 ± 0.02 vs 0.62 ± 0.01 mm, n = 64 vs 856/124, P < 0.0001 unpaired t test). The coefficient of variation in raw force data was much lower than in human right atrial trabeculae (0.532 vs 0.818, P < 0.001). In contrast to human atrial trabeculae, there was no positive correlation between cross-sectional area and force. As seen before for human right atrial trabeculae, normalization of force data to cross-sectional area increased data scattering (coefficients of variation was 0.956 vs 0.532, P < 0.001). That theoretical model was used to derive the values for fully oxygenated “active” area as a function of muscle radius for right atrial trabeculae (Figure 1C). While total cross-sectional area increased exponentially, the slope for fully oxygenated area increase was rather low and almost linear. The model predicted incomplete oxygenation for muscles with radius >0.2 mm. Regression lines for fully oxygenated area and for force generation run in parallel. The same holds true for guinea-pig papillary muscle (Figure 1D). To test whether normalization to fully oxygenated (active) area could reduce data scatter in right atrial trabeculae, we calculated coefficients of variation for data normalized to active area and compared to coefficients of variation in raw data and those normalized to total cross-sectional area. The coefficient for normalization on active area amounted to 0.832 and was significantly (P < 0.001) lower than for data normalized simply to total cross-sectional area. However, normalization to active area did not result in lower data scattering than using raw data (P = 0.414). For guinea-pig papillary muscles, the coefficient of variation was 0.665 when normalized to active area and again significantly lower compared to normalization to total cross-sectional area, but significantly higher than for raw data (P < 0.001). The fraction of fully oxygenated fibres in a superfused muscle preparation does not only depend on muscle thickness but also on actual oxygen demand.1 Any positive inotropic intervention that increases oxygen consumption should therefore increase the fraction of the muscle being hypoxic. Obviously, thicker muscles should be at higher risk for hypoxia when force is increased. To evaluate whether muscle thickness has an impact on effect size of a positive inotropic intervention, we have plotted the maximum positive response evoked by high concentrations of Ca2+ (8 mmol/L) as a function of muscle radius (Figure 1E). Unexpectedly, there was a weak, but clearly positive correlation between the inotropic effect of high calcium concentrations and muscles radius in right atrial trabeculae. We tested whether a much simpler approach could work better. By simply plotting stimulated (maximal) force vs basal force, we found a clear positive linear correlation with a rather high correlation coefficient of 0.6 (Figure 1F). Following the initial concept of not only Koch-Weser3 but also the more sophisticated computer model proposed by Han et al,1 limited oxygen supply should be a particularly critical factor when oxygen demand is increased, that is, under inotropic stimulation. Since we saw a rather flat curve for force vs area we have chosen the simplest approach and plotted delta force vs radius. As seen before for basal force there was a very low association between positive inotropic effect and radius. There was no hint that inotropic responses could be limited in thicker muscles. Again, R2 values were rather low. In conclusion, force values measured in intact muscle preparations differ widely. There is only a very weak correlation between force and cross-sectional area. Correlation with a predicted oxygenated area of muscles works better than normalization to cross-sectional area. However, cross-sectional area and oxygenated area do not contribute to more than 1 and 2% of variability in force. Other, not yet identified properties may contribute to variability. There are no competing interests in any of the authors.
No takes yet. Share an insight, caveat, or question.
Pecha et al. (2018) studied this question. Normalization of force to predicted oxygenated area vs. Normalization to total cross-sectional area or raw data was evaluated on Coefficient of variation in force data (p=<0.001). Normalization of force to predicted oxygenated area yielded a lower coefficient of variation than normalization to total cross-sectional area, but did not improve upon raw data.
Synapse has enriched 10 related papers. Consider them for comparative context: