1. The pathophysiology of chronic arm oedema after treatment of breast cancer was investigated by collecting serum and subcutaneous interstitial fluid from the affected and contralateral arms by the wick method (both arms) and by aspiration (oedematous arm). The fluids were analysed for total protein, albumin, glycosaminoglycan and viscosity, and arm volume was measured.
2. Total protein concentration in the aspirated oedema fluid was 32.4 + 7.5 g/l (mean + SD throughout; n = 39). Protein concentration in wick fluid from the oedematous arm (35.8 + 7.3 g/l, n = 14) was not significantly different from that in aspirated fluid. The oedema protein concentrations were significantly lower than in wick fluid from the non-swollen arm (41.4 + 6.7 cmH2O, n = 13, P <0.01, analysis of variance). This was surprising in view of the common assumption that, the condition being of lymphatic origin, the oedema protein concentration should be raised.
3. The ratio of aspirate protein concentration to serum protein concentration showed a weak but highly significant negative correlation with the percentage increase in arm volume (r = −0.47, n = 35, P <0.005), again in contrast to conventional expectation. The demonstration of a reduced protein concentration in the swollen arm did not therefore depend solely on a comparison with the wick control results. The volume increased by on average 33% and the ratio of aspirate protein concentration to serum protein concentration averaged 0.52 + 0.11 on the swollen side and 0.64 + 0.13 on the unaffected side.
4. Serum protein concentration in the patients with arm swelling (61.2 + 4.9 g/l) was significantly lower than that in postmastectomy patients without this complication (65.0 + 6.2 g/l). Most of the decrease occurred in the albumin fraction (oedema patients, 383 + 5.1 g/l; control patients, 42.0 + 2.1 g/l). In oedema patients receiving the anti-oestrogen tamoxifen serum albumin concentration was on average 23 g/l lower than in oedema patients not under medication (P <0.05, t-test).
5. Glycosaminoglycan concentration in oedema fluid was 0.8 +0.14 g/l (n = 21) and 75% was sulphated. Along with the plasma protein this raised the relative viscosity of the fluid to 1.34 + 0.16 (n = 11).
6. The reduction in interstitial protein concentration in the swollen arm, contrary to expectation in lymph-oedema, could be explained in several ways. One possible hypothesis in light of reported haemodynamic abnormalities in such arms is that capillary pressure rises, increasing capillary filtration rate. We conclude that the pathophysiology of postmastectomy oedema involves additional factors besides axillary node trauma, and we suggest that input (filtration) as well as output (lymph flow) requires evaluation.