Fluid exchange in the microcirculation

Starling's (1896) hypothesis, that the rate and direction of microvascular fluid exchange is proportional to local differences in hydrostatic pressure and effective colloidal osmotic pressure (or oncotic pressure) across the microvascular walls, has been so well supported by experimental evidence for so long (e.g. Landis, 1927; Pappenheimer & Soto-Rivera, 1948) that it can be called Starling's principle. The representation of Starling's hypothesis, however, as a balance of fluid loss from blood to tissues at the arterial end of the microcirculation and fluid uptake from tissues to blood at the venous end is supported by neither observation nor theory (Levick, 1991). This picture, which still appears in some textbooks, rests on the belief that mean capillary pressure equals plasma oncotic pressure and that hydrostatic and oncotic pressures of interstitial fluid are insignificant. In humans, mean capillary pressure approximates to plasma oncotic pressure only at or above heart level; the pressures in the majority of capillaries are considerably greater than this during waking life (Levick & Michel, 1978). Furthermore, when microvascular pressures do approximate to plasma oncotic pressure, the oncotic and hydrostatic pressures of interstitial fluid cannot be neglected.

Interest in the interstitial pressures was greatly stimulated by Guyton (1963) when he reported that interstitial hydrostatic pressure, P_i, had sub-atmospheric values in the subcutaneous tissues of a dog. He also showed that P_i changed with trans-capillary fluid movements, becoming more negative when fluid was absorbed from tissues and rising towards atmospheric pressure when microvascular filtration increased. These observations suggested that changes in interstitial fluid volume by capillary filtration and reabsorption might be buffered by changes of P_i. The mechanism seemed relevant to fluid exchange in the lungs where small changes in microvascular pressure give rise to transient filtration or re-absorption that rapidly attenuate. As an alternative mechanism accounting for this buffering of pulmonary fluid exchange, Lunde & Waaler (1969) suggested that it might be achieved by dilution and concentration of protein in a small interstitial compartment close to the pulmonary capillaries. Their idea received less attention at first and most of the work surrounded the negative values of P_i but by 1980, P_i was considered to be higher than first thought (but still negative in most tissues) and interstitial oncotic pressure was recognized as important for fluid exchange in most vascular beds (Aukland & Nicolaysen, 1981).

In the late 1960s and 1970s, my colleagues and I regularly confirmed Starling's principle while estimating the permeability coefficients of single perfused frog mesenteric capillaries (e.g. Michel et al. 1974). There were also puzzling observations. A step reduction in capillary pressure below perfusate oncotic pressure led to brisk fluid uptake but if pressure was held at this level, uptake attenuated and within a short time it could not be detected. It appeared that when capillary pressure was less than the oncotic pressure of the perfusate, it reset the balance of pressures across the vessel wall. If the interstitial pressures were being reset during capillary fluid uptake, it seemed more likely that the peri-capillary oncotic pressure was rising rather than P_i falling, since the mesentery was exposed and well hydrated with superfusate in these experiments. Although macromolecular permeability of the vessels was low, it was finite and we realized that, in a steady state, the concentration immediately outside a vessel would be determined by the ratio of the rate of macromolecular leakage through the vessel wall to the filtration rate. Uptake of fluid into a capillary should concentrate extra-vascular macromolecules, reducing the oncotic pressure difference across the wall and bringing fluid movement to a halt. These changes should reverse when capillary pressure and hence filtration rate are raised. Peri-capillary protein concentration should fall and the effective oncotic pressure difference across the vessel wall increase.

In 1976, I wrote out a simple expression incorporating protein permeability into the Starling principle for a talk on capillary fluid exchange. Evaluation suggested that it predicted our observations well and that it might be of general application. An improved version was finally published in a review several years later (Michel, 1984). Since it indicated that fluid uptake into microvessels cannot be sustained, it seemed unlikely that the textbook description of continuous fluid reabsorption in the venous microcirculation was possible. Continuous fluid uptake into venous microvessels was not supported by available measurements of microvascular and interstitial pressures but at the time, it was uncertain whether the direct estimates of microvascular pressure were representative (Michel, 1984). It seemed essential to obtain data that could be compared systematically with the theory. Mary Phillips and I measured transient and steady-state fluid exchange in single frog capillaries during the first few months of 1983. The results were very much as we had expected. Quite apart from demonstrating the differences between transient and steady-state fluid exchange, we were also able to make improved estimates of macromolecular permeability and reflection coefficient (Michel & Phillips, 1987). Our only surprise was how quickly a steady state could be achieved following a step reduction in pressure. It suggested that the steady-state protein concentration outside a capillary was established in a very small volume of peri-capillary fluid. Just how small it may be, became clearer 10 years later.

In an important analysis of microvascular fluid balance, Levick (1991) raised the question of why observed rates of lymph flow from various tissues were much less than those predicted from the microvascular and interstitial pressures. A possible explanation is that fluid and macromolecules cross microvascular walls by separate routes and the extra-vascular protein concentration that matters is that in the endothelial intercellular spaces between the glycocalyx and the tight junctions (Michel, 1997; Weinbaum, 1998; Hu & Weinbaum, 1999). Experimental evidence for this was obtained in frog mesenteric capillaries (Hu et al. 2000) and in rat microvessels (see paper by Adamson et al. in this issue of Journal of Physiology). The recent developments re-open many questions that were believed to be resolved 50 years ago. It will be interesting to see how well these problems are addressed in the years to come.

Original classic paper

The original paper by Michel & Philips published in The Journal of Physiology can be accessed online at: http://DOI: 10.1113/jphysiol.2004.063511

http://jp.physoc.org/cgi/content/full/jphysiol.2004.063511/DC1 This material can also be found at

http://www.blackwellpublishing.com/products/journals/suppmat/tjp/tjp246/tjp246sm.htm