The coast is cleaner.

Local free cell (eosinophils and macrophages) and extracellular (acid mucopolysaccharides) reactions were studied by histochemical techniques in 72 primary lung cancers (Group A), 17 pulmonary metastases and 5 lung tumours of unknown origin (Group B). Strong cellular reactions were more frequent in Group A than in Group B, whereas extracellular reactions were more frequent in Group B than in Group A. In both groups the degrees of eosinophilic infiltration and accumulation of acid mucopolysaccharides tended to be negatively correlated. In contrast, macrophages and eosinophils showed a positive correlation in the primary cancers, but none in the other tumours. The extracellular material had physical and chemical properties (solubility, affinity to stains and degradability by enzymes) of the sulphated acid mucopoly-saccharides. From its distribution it appeared to be derived from fibroblasts and/or tumour cells. IN some experimental tumour systems the host develops an immune reaction against a neoplastic transplant, mediated in part by infiltrating cells in contact with the tumour cells. Important effectors of this defence reaction appear to be macro-phages (Evans, 1972; Eccles & Alexander, 1974). On the other hand, the concept of "stromal reaction", an ill-defined term coined by Russell (1908) has emerged from time to time for various observations in human tumour pathology: for "inflam-matory infiltrates" in breast cancer (Berg, 1959) and carcinoma of the oesophagus (Takahashi, 1961) "elastosis" in breast carcinoma (Shivas & Douglas, 1972) "lymphocytic infiltrations" in gastric car-cinoma (Black et al., 1956), neuroblastoma (Lauder & Aherne, 1972) and ovarian cancer (Barber et al., 1975), macrophages in melanomas (Roubin et al., 1975) and plasma cells and antibody in lung cancer (Joachim et al., 1976). We now add another triad of pieces to the puzzle: eosinophils, macrophages and acid mucopolysacchar-ides in primary and secondary human lung cancers. Initially we were stimulated by the experimental "macrophage wave" but decided to focus on macrophages and eosinophils in the hope of gaining additional insight into the mechanisms of the host response. In the end the spectrum was wvidened by the inclusion of extra-cellular acid mucopolysaccharides. The choice of this parameter grew from accidental observations of odd strands of poorly structured metachromatic material between tumour cell groups and individual cells. No connection between the '3 parameters was implied at the beginning. A comparison between the results in 72 primary lung cancers, 17 pulmonary meta

IN some experimental tumour systems the host develops an immune reaction against a neoplastic transplant, mediated in part by infiltrating cells in contact with the tumour cells. Important effectors of this defence reaction appear to be macrophages (Evans, 1972;Eccles & Alexander, 1974). On the other hand, the concept of "stromal reaction", an ill-defined term coined by Russell (1908) has emerged from time to time for various observations in human tumour pathology: for "inflammatory infiltrates" in breast cancer (Berg, 1959) and carcinoma of the oesophagus (Takahashi, 1961) "elastosis" in breast carcinoma (Shivas & Douglas, 1972) "lymphocytic infiltrations" in gastric carcinoma (Black et al., 1956), neuroblastoma (Lauder & Aherne, 1972) and ovarian cancer (Barber et al., 1975), macrophages in melanomas (Roubin et al., 1975) and plasma cells and antibody in lung cancer (Joachim et al., 1976). We now add another triad of pieces to the puzzle: eosinophils, macrophages and acid mucopolysaccharides in primary and secondary human lung cancers. Initially we were stimulated by the experimental "macrophage wave" but decided to focus on macrophages and eosinophils in the hope of gaining additional insight into the mechanisms of the host response. In the end the spectrum was wvidened by the inclusion of extracellular acid mucopolysaccharides. The choice of this parameter grew from accidental observations of odd strands of poorly structured metachromatic material between tumour cell groups and individual cells. No connection between the '3 parameters was implied at the beginning. A comparison between the results in 72 primary lung cancers, 17 pulmonary meta-Correspondence to: I)r Editlh Kolb For differential staining of acid glycosaminoglycans the alcian blue (Scott & Dorling, 1965) and the acridine orange-CTAC (Saunders, 1964) methods were used, each wNith the proper critical electrolyte concentrations.

RESULTS
The histochemical observations were recorded wvithout knowledge of operative or other clinical data, and before the histological diagnosis was known. Initially the tumours were studied for infiltrating cells only, either eosinophils or macrophages or both. However, it became evident after 39 tumours had been examined, that "'metachromasia" observed with toluidine blue staining might also be worth systematic evaluation, because it was so conspicuous in some, but absent in other cases. Therefore, metachromasia was added as a thirdl parameter in the subsequent 92 tumours examined and herein presented. The incomplete results of the first 39 tumours were omitted.

Eosinophils
These cells, Awhich containi peroxidase in abundance, were easyT to identify by the brown-black reaction deposit of the enzyme developed histochemically. As Fig. I showAs, the cells were spread singly or in clusters in the scarce stroma of tumours with a compact structure. On the other hand, in tumours growing nmore invasively into the parenchyma of the lung, the eosinophils Awere more irregularly scattered and mostly localized in the interalveolar septa. In addition, the blood vessels nearby sometimes contained numerous eosinophils. Occasionally the adjacent parenchyma of the lung showed small groups of lymphocytes arranged in pseuidofollicles, sometimes surrounded by eosinophils.
Macrophages As Fig. 3 shows, these cells were identified by the reaction deposit for nonspecific esterases. Not every macrophage showed equally strong enzyme activity; all shades of intensities being observed (Fig. 4). The macrophages were distribuited in densely packed clusters wiithin septa and alveoli throutghout the parenchyma of the lung. They were, however, never detected within the venules surrouinding the tumour.

Mletachronasia
Some tumours, when staineld witlh alcian blue or toluidine blue, were interwovenwith intenselv metachromatic strands (Fig. 5) containing few cellular elements (Fig. 7). At higher magnification much metachromatic material could often be detected also between the tuimour cells themselves (Fig. 7). Alcian blue staining with the addition of different concentrations of magnesium chloride, allowed partial characterization of the mucopolysaccharides in the strands. At lowN concentrations of MgC12 (0.1-0 3Ai) the ground substance of the strandls showed nio staining, but the tuimouir cells were reactive (Fig. 5). With increasing concentrations of MgC12 (0.45, 0-6, 08M) the ground substance became more and more reactive, while the tumour cells lost the stain (Fig.  6). From these results the metachromatic material could be chondroitin sulphate, heparin or keratin sulphate, but not hyaluronic acid. Confirming results were obtained with the acridine orange CTACmethod.
Enzymic digestion studies showed the material to be resistant to hyaluronidase, neuraminidase, collagenase and trypsin; however, it could be washed out after digestion with pepsin. Therefore it appeared to be firmly bound to a protein backbone. Preliminary preparative extraction and electrophoresis (unpublished) showed that the metachromatic material had an electrophoretic mobility of its own, contained a small fraction of hyaluronidase-digestible, but a major nondigestible fraction. This fraction was also resistant to neuraminidase. Its electrophoretic mobility was distinct from extracted acid mucopolysaccharides of normal lung, trachea, cartilage, bronchial mucosa and peritracheal connective tissue used as controls, and from the markers commercial animal hyaluronic acid, chondroitin sulphates A and C and heparin. From these properties the material seemed to belong to the group of chondroitin sulphates B.

Grading of cellular and extracellular reactions
The following grading scales for the 3 parameters were used in the preliminary studies: (0), ( + ), ( + + ) and ( + + + ). In the definitive studies presented in this report, the degrees (+ + ) and (+ + +) were no longer differentiated. They were mentally combined and are designated (+ + + ) in the figures, meaning "strong reactivity" as opposed to "weak reactivity ( +)" and "no reactivity (0)". In Figs. 8-10 the reactions were correlated separately for the primary lung cancers and all the other tumours taken together. Shaded areas were used to point out the main parameter under consideration and its major relationship to one other parameter. Fig. 8   between macrophages and eosinophils. Numerous macrophages were observed in 45j70 primary lung cancers (A); in 30/45 they were accompanied by numerous eosinophils, in 11/45 by few, and in only 4/45 were there no eosinoplhils. No correlation between the 2 cell types was evident however in the other tumours (B). In contrast, an inverse correlation of a similar degree was found between metachromasia and eosinophils (Fig. 9). No metachromasia was found in 38/72 primary tumours (A); this was associated with strong eosinophilia in 24/38, few eosinophils in 12/38 and none in 2/38 cases. There was also an inverse correlation between the 2 parameters in the other tumours (B). Likewise, the correlation between inetachromasia and macrophages, as demonstrated in Fig. 1.0, tended to be negative in the primary lung cancers (A), but was n-ot clear-cut for the other group (B).

DISCUSSION
The identification of reactive cell types in human tumours is often neglected, mainly because the relevant enzymes are destroyed by fixation, the usual staining is inadequate for this purpose and surgeons are more interested in histology than in immunology. The results presented show that more than half of the primary lung cancers induced a strong infiltration by free cells, usually both macrophages and eosinophils, and only about a third showed strong metachromasia; on the other hand, locally strong eosinophilia wsas found in only 5/22 other tumours, but strong metachromasia in 10 of them. Thus, the 2 tumour groups showed opposite tendencies for the cellular and extracellular parameters, and within the same group, the 2 parameters seemed to be inversely correlated. However, strong metachromasia, present in 20 primary tumours, was not incompatible with strong cellular reactivity; about half of these tumours showed strong intensity for botlh free cells and metachromasia. Of the other tumours, 2/22 showed this type of reaction. These observations indicate: that patterns of local response can be discerned in malignant lung tumours; that cellular and extracellular reactions may be correlated; and that, there are differences between primary cancers (33 squamous-cell types, 9 undifferentiated tumours excluding oatcell types, 7 adenocarcinomas and 3 alveolar-cell carcinomas; for further details see Kolb & Mtiller, 1979) and metastatic tumours or tumours of unknown origin (II different histologies, including 6 hypernephromas and 5 teratomas). From the frequent pattern in primary tumours "strong cellular reactivity, no metachromasia" and the less frequent, but not to be ignored combination "strong cellular reaction, strong metachromasia", we suggest that the cellular and extracellular responses do not depend directly upon each other, but that the former expresses host responsiveness, while the latter expresses a complex interaction of VI az L-E I ;z 9 P. +++ o F~l+H~+fl IrVA + tumour cells and fibroblasts, with tumour cells perhaps inducing an abnormal fibroblast reaction (Dixon & Moore, 1953;Shivas & Douglas, 1972;Howard et al., 1976) possibly by producing acid mucopolysaccharides themselves (Takeuchi et al., 1976;Sampaio et al., 1977;Glimelius et al., 1978). However, strong metachromasia, frequently seen in metastatic tumours, might inhibit cellular responses. The possible significance of these observations will be evaluated in the subsequent paper by correlating them with clinical data (Kolb & MIuller, 1979).
Tlhis Mwork was supportc(l by a grant from the Hartmann MAaller-Stiftung.