THE ROLE OF SODIUM, POTASSIUM, CALCIUM, AND MAGNESIUM 1N CANCER: A REVIEW

M. J. SHEAR (From the Ofice of Cancer Inveeligationa, U. S. Public Health Service, Harvard Medicd School)

r. INTRODUCTION Among the many angles from which the cancer problem has been attacked is the study of the r81e of inorganic substances. A large number of the chemical elements, starting with aluminum and on through the alphabet to zinc, have been considered as possibly playing a part in tumor growth, and importance has been ascribed to some of them. Commou constituents of living tissue, as well as substances which are found in the body in very small amoynts, or not at all, are among those alleged to have significance in cancer. It is possible that some inorganic substance which is present in the body only in traces may ultimately be found to be of impor- tance in malignant disease ; on the other hand, investigations have also been directed towards determining whether one, or more, of the fixed bases which constitute so large a fraction of the total min- eral content of normal living matter plays a determining r61e in tumor growth. A voluminous literature abounds in reports assert- ing that the fixed bases of the body, i.e. sodium, potassium, calcium, and magnesium, exert a significant influence on neoplasms. A re- tarding effect on tumor growth has frequently been ascribed to calcium and to magnesium, while an accelerating effect has been ascribed to potassium and to sodium. Importance has also been attached to the ratio of one base to another, especially to the po- tassium : calcium ratio. Around this phase of the problem a large literature has grown up. Much of the work is comparatively recent. The first papers in this field appeared a quarter of a century ago, but most of the contributions have been published during the past decade. Since no comprehensive review of this subject exists,' the literature has been critically surveyed and the material assembled in a convenient form. Studies designed to throw light upon the part played by these four elements in cancer cover a wide variety of approaches and

1 Of the brief r6sum6s and partial reviews, the more rwent aro those of Lieber (1924), Kahn (1925), Bishop (1930), and Mankin (1932). 924 SODIUM, POTASSIUM, CALOIUX AND MAGNESIUM IN OANOER 925 deal with many aspects of the tumor problem. Thus, analyses have been reported for tumors and for the blood of tumor-bearing individuals. Attempts have been made to influence susceptibility to tumors and to affect tumor growth by treatment with salts of these metals. The substances have been administered in various ways: by mouth, by injection, by inclusion in the diet, and by im- mersion of tumors in salt solutions prior to inoculation. In addition to such direct attacks upon the problem, there have been a number of indirect approaches. For example, it has bee11 asserted that the calcium ion concentration of the blood is de- creased in cancer, and several investigators have administered parathyroid hormone in an attempt to increase this concentration. Vitamin D, which exerts an important effect upon calcium me- tabolism, has also been given to tumor-bearing animals. Furthermore, it is possible that these four bases may be in- volved in cancer metabolism in ways that have in the past received but little recognition. Sfagnesium, for example, is an activator of enqmes involved in carbohydrate metabolism, as von Euler (1931) recently pointed out. Warburg (1933) is of the opinion that in- organic salts may exert an important influence on living tissues “indirectly, through their influence on the oxidative reactions. ” Much study has been devoted to cell respiration and carbohydrate metabolism in cancer, but relatively little is known about the effect of the common bases on these processes in malignant disease. Critical sifting of the published experimental data reveals that our present knowledge of the part played by sodium, potassium, calcium, and magnesium in cancer is incomplete and unsatisfac- tory. A study of the literature leaves one with the impression that sodium is of no especial significance, that potassium may have a stimulating effect, and that calcium appears to have a retarding effect on tumor growth. No substantial evidence is found for the oft-repeated assertion that magnesium has a beneficial effect in cancer. These are the indications furnished by the existing evi- dence. They are not, however, convincingly established concln- sions, of which there are but few. Apart from its academic interest, this problem has a direct bearing on clinical practice. For example, one school has been holding magnesium deficiency responsible for cancer and accord- ingly has strongly recommended administration of magnesium salts not only in treatment but also in prophylaxis, while the re- sults of other investigations indicate that magnesium may actually stimulate tumor growth instead of retarding it. Similarly, an- other school accepts the thesis that a deficiency of calcium ions is responsible for tumor growth and recommends administration of parathyroid hormone therapeutically and prophylactically, while 926 M. J. SHEAR other work indicates that administration of parathyroid hormone stimulates tumor growth. In this survey the numerous contributions to the literature have been correlated and evaluated with a view to determining which of the conflicting conclusions rest on the soundest basis, and to ascer- taining which direction further investigations may profitably take. 11. TUMORANALYSES Sodium, Potassium, Calcium, and Magnesium Contents: The analyses of Beebe (190445) and of Clowes and Frisbie (1905) laid the groundwork of our present knowledge of the inorganic compo- sition of tumor tissue. In Beebe’s pioneer contribution it was re- ported that fresh, vigorously growing human tumors were richer in potassium and poorer in calcium than degenerating tumors. Clowes and Frisbie reported similar results for mouse tumors. Beebe analyzed the tumors for sodium, potassium, and calcium and gave the data in terms of the dry weight of the tissues. The amount of magnesium found was so small that quantitative deter- minations could not be made. Calcium was precipitated as oxalate and weighed as the oxide. The sodium and potassium were weighed together as sulfates ; the potassium was then precipitated as the chloroplatinate, and the sodium obtained by difference. Thus, the sodium content waa not obtained by direct analysis, but by’ calculation. In preparing the tumors for analysis, the tissues were ashed at dull red heat. Only 8 malignant tumors were analyzed by Beebe, and these tu- mors (human) were of various types. Clowes and Frisbie analyzed 100 mouse tumors, all adenocar- oinomas (Jensen). In a systematic investigation they correlated the characteristics of this tumor with the potassium, sodium, and calcium contents. Magnesium was never found in appreciable quantities. They found that “the amounts of potassium and cal- cium are subject to most remarkable variations ; rapidly growing, large tumors showing a marked tendency to a high potassium and low calcium content, whilst the reverse is to be observed in the case of slow-growing tumors. This is so marked that in certain cases of very early tumors potassium only is to be recognized, whilst in the majority of old, necrotic tumors calcium only is present. It will be seen, furthermore, that a steady fall from a high percentage of potassium in the younger to practically none in the older tu- mors is accompanied by a corresponding increase from no calcium in the younger, to a high percentage of calcium in the older, ne- crotic tumors. ”’ No mention wag made as to the analytical methods employed: neither was it stated whether the tissues were prepared for analysis SODIUM, POTASSIUM, CALCIUM AND MAGNESIUM IN CANCER 927 by incineration or by digestion with oxidizing solutions. The data were given in terms of the original wet tissue and since no values for the H,O content were reported, the results cannot be recalcu- lated on the dry basis. Sixteen years elapsed before the next important contribution, that of Waterman (1921), appeared. The work of Beebe and of Clowes and Frisbie had stimulated a number of investigators to study the rSle of potassium and of calcium in cancer, and for the most part subsequent students accepted the results of Beebe and of Clowes and Frisbie quite uncritically. Thus Neuberg (1911), in his review, stated that Beebe had carried out an exhaustive in- vestigation on the mineral content of human tumors. As a matter of fact, Beebe’s work was far from being exhaustive: indeed, he himself had said: “It is impossible from so few experiments to draw definite conclusions as to the function of these metals. . . .9, Waterman felt that it was desirable to study this problem anew in order to confirm, if possible, these results which had been so un- critically accepted. He therefore proceeded to analyze carefully a series of about 30 assorted tumors of man, dog, rat, and fowl. Analyses were made for sodium, potassium, and calcium; the data were given in terms of dry weight of tissue. In order to avoid the loss of potassium which may occur during the ashing of tissues, Waterman destroyed the organic material ac- oording to one of the 80-called “wet-ashing” methods, i.e. by diges- tion with nitric and sulfuric acids. Although it has been recog- nized for some time that dry ashing may result in serious losses of certain inorganic constituents, this point has not been taken into consideration in many investigations on the composition of tissues. As regards the accurate determination of potassium, Waterman stated that there was no method available which was above criti- cism. He used both the standard chloroplatinate method and the then new cobalti-nitrite method of Kramer and Tisdall (1921~). Sodium he tried to determine directly as pyroantimonate, using the method of Kramer and Tisdall (1921b),which had ‘ust been de- vised. He could not get a pure reagent, however, and poor results were obtained. Consequently the procedure of subtracting the value of KCl from the total chIorides was followed. Thus, like hi8 predecessors, Waterman obtained sodium bv difference and not by actual analysis. He confirmed the general conclusions of Reebe and of Clowes and Frisbie and summarized his findings as follows : A. The more rapid the growth of the tumor, the greater is the potassium content. B. The older the tumor, or portion of a tumor, and the slower the growth, the less its potassium content and the greater its cal- cium content. 44 028 M. 3. sHl4AR C. There is an antagonism between sodium and potassium in that the more rapid the growth, the greater is the proportion of potassium. D. In general, an increased potassium : calcium ratio indicates a rapidly growing tumor without degenerative ohanges. Waterman examined a series of Rous fowl sarcomas in various stages of growth and degeneration, Old degenerating tumors showed a lower potassium content and an increased calcium oon- tent, whereas in fresh metastaaes from the same animal these find- ings were reversed. In a later paper Waterman (1922) repeated his oondusions regarding the relationship between the potassium : calcium ratio and malignity. In addition to these contributions, a number of studies of lesser importanm were reported prior to the last decade. Mottram (1910) analyzed 19 human tumors for sodium and po- tassium and found wide variations in both constituents. The method employed was crude, the amount of alkali being determined by the length of time it oould be detected in the flame in a spectro- soopia examination. The findings were not in accord with those of Beebe, but inasmuch as his method was only roughly quantitative, Mottram’s results cannot be taken as invalidating those obtained by quantitative chemical analyses. NBgre (1910) reported values for the sodium doride and po- tassium chloride contents of 2 transplanted mouse tumors. He stated that these potassium results were not in word with those of Beebe, but it is obvious that little aignifioance can bb attached to these isolated analyses. No mention was made of the methods em- ployed by the analyst, Piettre, whose results NPgre reported. Robin published a series of papers which are frequently oited. In one article (1912) he reported analyses of human livers for sodium, potassium, magnesium, and calcium. Only one cancerous liver was analyzed. A second pager (1913a) repeated the data previously published and induded a study of one more cancerous liver. In a third paper (191%) analyses were reported for the canoerous and healthy parts of a few livers ; the conolusions drawn regarding magnesium, sodium and potassium as “Phcipes ett esc8s” and calcium and iron as 6‘P&u$pes d6ficitaires” in cancer tissue are not warranted by the data. A dmade later Robin (1922) reported a few more analyses of hepatic oancers in man. He treated 3 patients with a mixture of oalcium glycerophoaphate, magnesium hydroxide, and magnesium, sodium and potassium silicates. He later analyzed cancerous tissues (one cancer of the colon, one of the esophagus, and one of the liver) for silica, wlcium, and magnesium. The results were oompared with those obtained on analysis of 2 cancerous livers of patients who had not received this medication and with the re- sults from a normal liver or livers. The average values only were given; the individual variations within each group therefore can- not be compared with the differences between the groups. More- over, the number of normal livers analyzed was not stated, while the number of cases was so small that no generalizations could safely be made. Nevertheless, Robin stated that his analyses indi- cated that cancerous tissues fix silica, lime, and, to a less degree, magnesium. He furthermore recommended the use of silica com- pounds in the treatment of cancer I During the last decade, contributions to this subject have been more numerous. Bohdenburg and Krehbiel conducted several se- ries of tumor analyses. They (1922) analyzed rat tumors for sodium, potassium, and calcium and compared the results with those obtained on analysis of the testes and livers of the tumor- bearing rats and of normal rats. The methods used were those of Kramer and Tisdall. Bix analyses were reported for each of the following : growing Jensen sarcoma, receding Jensen sarcoma, and spontaneous sarcoma of the liver. The values obtained for each of the three constituents varied rather widely. The potassium con- tent and also the calcium content averaged less in receding than in growing tumors. No mention was made as to which method of preparing the tissues for analysis was employed. The results were given in terms of wet tissue, and no values were reported for H,O content. In a continuation of this work, Bohdenburg and Krehbiel (19%) reported analyses for six types of transplantable rat tumors. The analytical methods employed were the same as described in the earlier paper. Five types of sarcoma and the Flexner-Jobling carcinoma were analyzed for sodium, potassium, and calcium. From five to seven analyses were given for each type of tumor. Here, too, the results were given in terms of wet tissue. The variations in the sodium and potassium contents of each type of tumor were rather large. This, together with the fact that the water content of the tumors was not given, raises a question as to the significance of the differences between the average values for the various types of tumors. According to the authors, there ap- pears to be a rough parallelism between the average content of po- tassium of a given type of tumor and the percentage of successful takes on transplantation. These same investigators studied the effect of x-rays on the mineral aontent of rat tumors. They found that “rat sarcomata subjected to lethal doses of radiation show, twenty-four hours after exposure, hypermineralization with negligible changes in the per- oentages of the salts determined and seventy-two hours after 930 M. J. SBUB radiation, a reduction of the salt oontent with an increase in the relative amount of sodium.” Analyaes were reported for potae- aim, sodium, and aaloium contents of three types of rat sarcoma and two groups of controls; from three to six analyses were given for ewh group. However, here, too, the variations within each group were rather large. When the large average deviations are taken into amount, the differences between the means do not ap- pear to be so striking. Wolf (1923) stated that the ratio of potaasim to calcium in oanoer tissue was roughly parallel to the intensity of proliferation, varying between 15 and 20 for rapidly growing tumors and between 5 and 10 for slowly growing tumors. He worked with a mammary epithelioma (mouse). In normal mouse tissue this ratio varied from 5.4 (liver) to 0.05 (oonnmtive tissue). However, wide di- vergencies were noted in cancer tissue, values as low as 0.5 being obtained. Wolf did not give any detailed data. The values for ornloinm and potassium were not given, nor the number of tumors analyaed. It is, therefore, impossible to decide how much impor- tanoe is to be attached to the conclusions or to his reoommendation that oaloium ions be employed in cancer therapy. Roffo (1924-250) also analyzed normal and tumor tissues for potassium and oalcium, presenting his data in nine tables. The material was prepared for analysis by digecltion with nitric and sulfuric aoids. The potassium was determined a8 the cobalti- nitrite and the calcium according to the Kramer-Tisdall (19216) method. The following tissues were analyzed: chicken embryo from before fertiliaation to eighteen davs; chicks up to eighteen days of age ; normal white rats from birth to six months : rat sar- omas and carcinomas of varioue ages, and normal tissues of tumor-bearing rats. The weights of the tissues were given both wet and dry. It was found that the potassium content of the tumors was about twiee that of the rest of the rat, and about the same as that of rat embryos ; importance was therefore aacribed to potassium in proliferating tissue. 8ince the values obtained for calcium did not exhibit any regularity, no conclusions were drawn regarding it8 r81e in tissue growth. The work of Loeper, Turpin and Zizine (1925) is frequently aited as demonstrating a relation between the development of a tu- mor and its potassium content. These investigators analyzed grafted mammary epitheliomas of mice for potassium and sodium. They used a method of “wet-ashing” for preparation of the tissues and employed the methods of Kramer and Tisdall in the analyses. They state that castration diminishes the potassium content of the body in mice; that the potassium content of grafts in castrated ~OD~M,POTANHUM, ols~mxAND MA~NEIIUM IN OANOEB 931 mice ie much smaller than that of grafts in control mice ; and that following castration the tumor grafts develop abnormally and fre- quently regress. Inspection of their scanty data, however, shows that the potassium content of the entire body was found to be 0.26 per cent and 0.27 per cent in castrated mice, and 0.35 per cent and 0.26 per cent in non-castrated mice. To conclude frm such evi- dence that castration diminishes the potassium content of the body ie obviously unwarranted. Moreover, only one value for potas- sium was given for each type of tumor. Butts, Huff, and Palmer (1927) ashed normal and tumor tissues obtained from various souroes and determined the mineral consti- tuents by means of the epectrograph. They state that tumor tissue contained muah more sodium than did normal tissue. Since, how- ever, the tissues were "ignited in a platinum crucible and thor- oughly ashed in the flame of a blast lamp" and the spectrographic exaidnations were made on the ash, it is, of course, impossible to know how much of the sodium was lost during ignition in the blast. The findings of Goldfeder (1928) were contrary to those ob- tained by previous analysts in that little difference was observed in the calcium content of young and of old tumors. She reported 11 analyaes of transplantable mouse carcinomas ; the calcium content was remarkably constant, varying between 0.0026 and 0.0039, with an average of 0.0031 per cent. Another series of mice was divided into two groups of 5 each; one group was analyzed when the car- cinomas were about the size of a walnut, the other group when the mibe began to die. The calcium content was found to be the same in both groups. ooldfeder also reported that oral administration of CaCl, caused an increase in the calcium content of these mouse carcinomas. In 21 analyses of control tumors, the calcium content was found to vary but little ; the limits of variation were 0.0021 and 0.0039. In the mice which had been treated with Cam, the results also showed but little variation; the extreme values (12 analyses) were 0.0029 and 0.0042 per cent, with an average of 0.0038 per cent as compared with an average of 0.0030 per cent in the control mice. (The value of 0.0404 "/OO which Goldfeder gave in the text and the value of 0.0094 which she gave, misplaced, in the table as the average for the treated mice are both typographical errors.) The difference be- tween 0,0038 and 0.0030, although not large, may be significant in the light of the individual 12 and 21 analyses, reapectively, of which they are the averages. aoldfeder conducted analogous experiments with Bous chioken s'srcoma inoculated into the legs of hens. The calcium content of the leg muscle of healthy hens was found to be remarkably con- stant: 4 analyses varied between 0.0065 and 0.0079 per cent, with 932 Y. J. BEEAB an average of 0.0073 per cent. The calcium content of the healthy leg muscles of 4 hene bearing thirteen-day-old tumors vaned from 0.0044 to 0.0063 per cent With an average of 0.0056 per cent. A group of 4 hens bearing twenty-threeday-old tumors gave similar values for the oalt5um content of healthy leg muscle; the values ranged from 0.0048 to 0.0060 per cent, with an average of 0.0054 per cent. The calcium content of the thirteen- and twenty-three- day-old tumors was also remarkably constant: the extreme limits of variation of 16 analyses were 0.0021 and 0.0040 per cent. The averages of these 4 groups. of tumors (4 in each group) were 0.0029, 0.0032, 0.0034 and 0.0030 per cent. The data are sum- marized in Table I. It is thus seen that tumors implanted into leg muscles of hens had a lower calcium content than healthy leg muscles of the same tumor-bearing hens. Furthermore, the healthy leg muscles of tumor-bearing hens had a lower calcium content than the leg muscles of control, non-tumor-bearing hens.

TAB= 1: Average vduS8 fw Ghbken 8tW60W (Goldfeder, 1088)

hlhd The aalaium Content Healthy control hens Healthy lag mde 0.0078 per cant Unreated tamor-bearing henr Healthy leg mde 0.0066 (I 61 66 66 66 I6 66 0.0064 6I 66 I6 61 Tumor (necrotia area) o.ooao 16 II 16 Id I1 66 dl 0.0034 I4 66 I6 66 Tumor (half-neorotia area) 0.0082 66 66 66 46 66 46 (I 44 0.0080 CaUl,-treated tumor-bearing hens Tumor (half-necrotic area) 0.0068 I6 66 (I , I6 66 Id I6 66 6I 0.0048 66 66 16 66 66 Tumor (necrotic area) o.oo5a I1 66 66 66 66 66 I6 66 0.0060 66 4I I4 I6 I6 Healthy part of leg mfwb 0.0068 61 61 66 66 66 66 66 4* 66 66 0.0080

To 8 hens bearing the sarcoma, CaCl, was administered orally. This caused a pronounced increase in the calcium content of the tumors and a smaller, but definite, increase in the calcium content of the healthy leg muscles of the tumor-bearing hens. The values for both tumors and healthy leg muscles were raised significantly towards the value for the leg muscle of healthy control hens. The neorotic and partly necrotic areas of these tumors were analyzed for oalcium separately; the results obtained were all fairly close together. There was little difference between the necrotic as compared with the partly necrotic areas ;there was also little difference, as regards the calcium content, between the thir- teen- and the twenty-three-day-old tumors. This checked with the results obtained for calcium in twelve- and twenty-one-day-old mouse carcinomas. In several respects, therefore, Goldfeder's results do not agree with current views regarding the increased calcium content of tu- mors with increasing age or with increasing extent of necrosis. Goldfeder prepared the samples for analysis by incineration. The ash was taken up in HC1, and the phosphate was removed by double precipitation with iron acetate. Caldum was determined on the combined fitrates by precipitation as oxalate in alkaline solution. *The results were apparently given on the wet basis. lzhis paper is a report of an extensive investigation. The pub- lished data constitute only part of the results obtained, but are said to be typical of all the data obtained. Analyses of Flexner rat carcinoma were made by Kimura (1928); mmparative analyses were performed at the same time on the liver and muscle of the tumor-bearing rats. Analyses were made for a number of inorganic constituents including sodium, po- tassium, calcium, and magnesium. Tumor material was analyzed twenty and forty days after transplanting; forty-day-old tumor tissue was also analyzed after carefully dissecting it free from soft material. However, since only one set of analyses was reported for each of these 3 types of material, no general conclusions may be drawn from these data. Xt wae found that tumor tissue con- tained more sodium, about the same amount of potassium, and about the same amount of magnesium as the two normal tissues, for emh of which two sets of analyses were given. The whole tu- mors uontained more calcium than the normal tissues, whereaR the firxu part of the tumor contained about half that of the normal tissuei Goerner and 8Mroff (1928) determined the calcium content of Flexner rat carcinoma in 4 control rats and in 9 rats that had re- ceived parathormone. In the treated animals the calcium content of the tumor was double that of the controls. In an analogous ex- periment with 6 test and 5 control animals bearing sarcoma 10, treatment with parathormone did not affect the calcium content of the tumors. Rabbit tissues were analyzed for potassium and cali?ium by Nagaoka, Hirata and Fukunaga (1929), who made a comparative study of sarcomas and of tissues in which infiRmmRtion had been produd by various irritants. An increased content of potassium and a deereased content of calcium were reported for both types of pathological tissue. These changes were stated to be more pro- nounced in tumors than in inflamed tissue. The necrotic material of tumors was found to contain less potassium and more calcium than the non-necrotic portions of the same tumors. The authors stated that their work corroborated that of earlier workers, who found that rapidly growing tumors contain more potassium than slowly growing ones, that there is some relationship between PO- 984 M. J. SEHAB tassinm content and cell groyth, and that the effect of calcium in this respect is antagonistic to that of potassium. The single table of data presented by Nagaoka and his asso- ciates gives the water content of the tiesues as well as the potas- sinm'and calcium values. However, analyses were given for only 5 tumors, and in only one wee was the necrotic material analyzed as well as the non-necrotic portion. The authors reported that high potassium-calcium values were obtained for rapidly growing tumors and low values for slowly growing or degenerating tumors. Although these results were said to be in agreement with those of previous workers, the number of tumors studied was muoh too small to establish convincingly any relationship between rate of growth and chemical compoaition. Bolaffi (19304 analyaed transplanted adenocarcinomas of mice which had been treated with magnesium chloride. Analyses were made for potassium, calcium, and magneaium after bringing the tumors into solution by treatment with nitric and sulfuri~mids. The alcoholic CaSO, method was used for calcium; potassiqm was determined as the cobalti-nitrite and magnesium was weighed as the pyrophosphate. The results were given on the wet basis. No evidence of awumulation of magnesium in the tumors WBB obtained. There was a diminution of the potassium : cddum ratio in a few instances, dueohiefly to increased calcium content, but this diminution was not oonfined to the cases showing diminished rate of growth. Seven analyses were reported for tumors from treated mioe and 5 analyses for tumors of control mice. In a subsequent paper, Bolaffi (1930b) analyzed tumors of mice that had been treated with other magnesium salts and tumors of control mice. Four analyses were reported for caloium and po- tassium. As in the preceding study, the number of analysea per- formed was too small to permit the differenaes obtained to be re- garded as significant. In studying the lipoid constituents of Ehrlich mouse adeno- carcinoma, Bolaffi (1931) noted that potassium soaps were par- ticularly abundant in the extracts from tumor tisaue and from other tissues of the cancerous animal, whereas almost negligible amounts were found in the tissues of the normal animal. Buchwald (1930) analyzed Jensen rat sarcomas and both spon- taneous and transplanted mammary carcinomas of ~cefor cal- cium and for the various forms of phosphorus. Some normal rat tiseues were also analyzed. Calcium was determined according to Rothwell's (1927) method. Six analyses were made of the trane- planted mouse tumor, 3 of the spontaneous mouse tumor, and 6 of the Jensen rat sarcoma. No attempt was made to correlate the calcium content With the malignancy or the rate of growth of the SODIUM, POTdBBmY, OALCIXUM AXD MAGNESIUM IN OANOEB 938 tnmors. It was found that both the calcium and orthophosphate contents of the tumors were low when necrotic material was ex- cluded from the tissues analyzed and that both were often very high when the necrotic areas could not be completely removed. This parallelism between orthophosphate and wlcium contents indi- catee, Buohwald believed, that necrotic areas contain considerable amounts of calcium phosphate. Magath and Kolomijets (1930) studied the calcium aontent of x-rayed and untreated tumors (Ehrlich mouse carcinoma and Jensen rat sarcoma). They also analyzed normal livers. It was found that the calcium content of normal livers and of tumors varies rather widely from one individual to another, and that it is a diffiaultmatter to get tumor tissue completely free from necrosis. Since the necrotic areas are very rich in calcium as compared to healthy tumor tissue, these investigators took pains to obtain, for analysis, material that was not necrotic. The tissues were brought into solution by treatment with nitric acid and perhydrol. Magath and Kolomijetz stated that exposure to x-rays did not produce an appreciable increase in the calcium content of the tu- mors. Although these careful workers give 5 tables of data in this paper, the data on which this particular conclusion was based are not given. A brief but important paper by Eichholtz (1931) contributes valuable information. He inoculated rats with a rapidly growing ~amomawhich, after eight or ten days, he removed under anes- thesia. Beveral implants were made in various parts of the ab- domen of each of 7 rats. 'Phe second tumor was removed several days after removal of the first. The tumors were prepared for analysis in a careful fashion: the active, non-necrotic areas were freed from blood by means of blotting-paper, and were brought into solution by treatment with nitric acid and perhydrol. Cal- cium was determined according to Kramer and Tisdall, and the magnesium according to the hpdroxvqninoline method of Eichholtz and Berg (1930). At least two portions of each tumor were analyzed separately; in most cases three, and in a few instances four areas of each tumor were analvzed separately. Furthermore, dnplimte determinations were made on each portion of tumor. ~eaalciumcontent varied between 70 and 367 mg. per 100 gm. of dwtnmor tissue. Tn spite of these wide variations from tumor to tumor, the calcium content of different portions of the same tu- mor did not vary more than about 2 per cent. Of special interest is the finding that the different tumors in the same animal had the same calcium content. Eichholtz also analyzed tumors for magnesium. The literature contains few such analyses. BeeBe, and Clowes and Frisbie, stated 936 M. J. BHBUB that the amount of magnesium was too small to be determined; Robin reported on the magnesium content of a few cancerous livers, and Bolafli on the majpesium content of a few specimens of mouse adenocarcinoma. Eichholtz worked out a new method for determining small amounts of magnesium and found that the type of rat sarcoma which he used contained significant amounts of magnesium. Analyses of 2 tumors are given: 2 portions of each tumor were analyzed separately, and each portion was analyzed for magnesium in triplicate. These tumors contained about 360 mg. calcium and about 125 mg. magnesium per 100 gm. dry sub- stance. Another excellent paper, by Kluge (1931 ), has been contributed from the same laboratory, the Pharmacological Institute of the University of Khnigaberg. In a brief article Kluge reported on the storage of calcium compounds in sarcomas. She implanted as many as 7 sarcomas in the same rat, using an extension of the technic employed by Eichholtz. After ten to eighteen days, cal- cium salts were injected subcutaneously and the calcium content of the tumors was determined. The tumors were removed at hourly intervals. In the control rats, one tumor was removed before the injwtion and the other from one to six hours after the injection. The calcium salts studied were CaCl,*6H,O, calcium gluconate and calcium catecholdisulfonate. The latter salt had been Rhown by Heaht and Eichholtz (1929) to arrest glycolyeis in tumors ivi vitro. Duplicate sets of experiments showed that injection of Cam, caused a much greater increase in the calcium content of tumors of rats anesthetized with avertin than in nnanesthetized rats. The use of anesthesia was therefore discontinued. After injection of CaCI, the calcium content of the tumors rose rapidly, attained a maximum in two hours, and then decreased slowly until, after six hours, it had about the same value a6 at the beginning of the experiment. At the end of two hours the increase averaged 48 mg. per 100 gm. dry substance: the individual in- creases (5 determinations) varied from 40.5 to 66.7 mg. per 100 gm. dry material. With calcium gluconate a similar type of curve was obtained, i.e. the calcium content rose to a maximum in two hours and re- turned at the end of six hours to the original value. The maximnm rise, however, averaged only 29 mg. per 100 m. dry tissue; the individual increases (4 determinations) ranged from 23.3 to 37.2 mg. ner 100 gm. dry tumor. When the other orRanic calcium compound was used, a barely perceptible increase was obtained. Here the maximum increase (6.7 mg. per 100 gm. dry tumor) waa noted at the end of three hours : the individual increases' (4 determinations) ranged from 5.3 to 7.5 mg. per 100 gm. dry tissue. SODmM, POl'asSma6, 0-Af AED ASAGNEBIUM IN OANCWB 937 Several important conclusions may be drawn from Kluge's fine pieoe of work. First, it shows clearly that calcium can rapidly penetrate into tumor tissue and that it can be taken up by the tu- mor in considerable amounts. In the next place it shows that the added calcium is not retained by the tumor but that it is rapidly re- turned to the host. Finally, the form in which calcium is injected determines whether it will be taken up by the tumor. It was found that inorganic calcium is absorbed by tumor tissue while organi- cally bound calcium is taken up slightly or not at all. Since the work of Koviioa and Ambrus (1931) was not available in the original, only their conclusions can be given here. They found that the potassium content is always high in malignant tu- mors, but is normal in benign or still incipient tumors. The aver- age potassium of healthy tissue is given as 1.4 per cent and that of malignant tumors as 2.3 per cent. The method of Kramer and Tisdall was used. Hoffmann (1931) reported that bone marrow contains a large amount of potassium which diminishes to about one-fourth at the conclusion of growth activity, but that in individuals bearing car- cinomas the potassium content is increased, being at least twice as great as in normal adult individuals. The bone marrow was ashed, and the ash heated to constant weight, but no mention is made of controlling tHe temperature to avoid possible loss of salts by volatihation. Calcium was determined as CaSO, ;sodium and po- tassium were determined together as mixed chlorides; the potas- sium was precipitated as K,PtCl,, and weighed as Pt. The sodium was obtained by difference. Analyses were done in duplicate in many cases. Values are given in terms of per cent of ash. The niaterial analyzed consisted of bone marrow from a calf, an ox, a five-year-old child, three non-cancerous men, and 6 patients with neoplastic disease. In a patient with sarcoma the bone mar- row contained the same amount of potassium as that of the non- cancerous adults. In the 5 carcinomatous individuals, however, the potassium values were more than doubled. In an individual with a skii carcinoma, the marrow of the tibia of the affected limb contained 4.94 per cent potassium. After the tumor was removed, the same bone showed 1.76 per cent, while the healthy tibia gave 0.83 per cent. Hofhann concluded that the changes in the sodium content, while also large, were not characterized by a reelarity similar to that of potassium; the changes in calcium content were not cor- relatedwith the growth process. It may be pointed out that the number of noncancerous individuals of various ages that were studied was too small to serve as a basis for correlating potassium oontent of bone marrow with growth. 938 M. s. 8- Histologicd Detection of Bodium, Potasskm, Calcium, mad Mapesium: In an attempt to supplement the findings of Beebe regarding the increased potassium content of tumor tissue, Tracy (1906-6) used Maoallum’s (1905) method for 1ocaliP;ing potassium in microscopic sections, but the method was found to be unsatis- factory. Tracy says: “In my experiment6 I have endeavored to follow in every particular the procedure adopted by Macallum, but I have not been able to confirm his results. . . . Until, however, fhe future reveals some more satisfactory procedure for separating the reagent from the triple salt formed, and until Macallurn’s work can be confirmed by others, the method appears not sutllciently re- liable to be of any real value.” Cattley (1907), who also used Maaallnm’s method, came to the conclusion that “Potassium is present in greater amount in the cell cytoplasm of the more actively growing portions of a malignant tumour than in older and less rapidly growing portions.’’ He sug- gested that Traay may have failed to differentiate between the two effects of the cobalti-nitrite reagent : “the staining action, which affects both cytoplasm and the nucleus, and the precipitating ac- tion which affects only potassium. When the nucleus is deeply stained an appearance is not infrequently produced which alosely resembles that of the potassium precipitate. ” Fifteen years later Bohdenburg and Krehbiel (1922) attempted to localize a large number of constituents of the cell by means of staining or by micro-precipitation tests. Among others, results were reported for potassium, sodium, and calcium. For potassium they used Macallurn’s method of precipitation with cobalti-nitrite followed by treatment with ammonium sulfide. With this technic, the orange-red preoipitate, obtained on addition of the cobalti- nitrite, turns black on treatment with sulfide. For detection of calcium they also used Macallurn’s method, which consists of plaa- ing the section in snlfuric acid, followed bv treatment with lead acetate and then with ammonium sulfide. The first precipitate of CaSO, is thus changed to PbSO,, which in turn is changed to black PbS. The sodium was detected by a modification of the Kramer- Tisdall method for sodium in blood: the section is placed in the pyroantimonate solution and the sodium is precipitated in crystal- line farm. Rohdenburg and Krehbiel summarized their Andings as follows : “Calcium is demonstrated in the nucleus as a heneedle-like precipitate, while that in the cytoplaem is finelv granular. Large coarse pranules are found between the cells. The ratio of calcinm within the cell to that without the cell is approximately as 1to 20, “Potassium is demonstrable in the cvtoplasm of the cell as a granular deposit, but is absent in the nucleus. The ratio of potas- aium within the cell to that outside the cell approximates 1 to 5. SODIUM, POTASSIUM, UALOIUY AND MAQNIOSIUM IN UANOIOB 989 “Sodium may be demonstrated as a crystalline deposit in the spaces between the cells and in the nucleus. When the cell is de- generating it is found in the cytoplasm.” The resulte obtained by such methods are of doubtful value. Rohdenburg and Krehbiel realized this. They wrote: “While the technical methods employed precipitated the various elements in definite places in the cell, it does not seem reasonable to suppose that during the life of the cell these elements exist in the manner present in the microscopic preparations. It is more probable that they exist in solution, which in turn presupposes a certain degree of ionization or of protein combination. A oareful comparison of many fields and many slides suggested variations in the potassinm, sodium, and calcium content of tumors in various phases of their groivth, though it is needless to say that the crude methods em- ployed warrant no definite conclusions.” Rohdenburg and Krehbiel (1924) employed these same micro- chemical reactions in a subsequent study in which they investigated the effect of x-ray treatment on the sodium, potassium, and calcium contents of tumors. They prepared sections in this way of a Flexner-Jobling oarcinoma before radiation and twenty-f our, forty-eight, and seventy-two hours after radiation with a lethal dose. Camera lucida drawings of these sections are given in their paper. The authors did not state what conclusions they drew from them ewtions ;the drawings are not accompanied by any comment. Waterman (1922) studied the histologic distribution of potas- sium and calcium in tumor tissue, employing Hamburger’s (1915) modifioation of the cobalti-nitrite reagent for potassium and Macal- lum’s method for calcium. He attempted to link his histological observations with the chemical analyses of tumors and with the physico-ohemical studies he had conducted on tumor tissue. Macallum’s technic was also used by Wolf (1923), who ex- amined sections of mammary epithelioma of mice and reported that the calcium was distributed uniformly and diffusely, while the potaeJium was found in the vicinity of the nucleus. hdreyt (1923) used staining methods for the detection of a number of constituents of tumor tissue (sarcomas). Among the elements studied were potassium and calcium. He used sodium aobalti-nitrite to detect potassium and silver nitrate to detect cal- dam. He concluded that the potassium precipitate increased in importance with increasing activity of the malignant cell, and that the calcium is more abundant when the sarcomatous cells are less active. The criticism which Ruhdenburg and Krehbiel levelled at such histological work is applicable here. Furthermore, silver nitrate is not a test for calcium; the recent work of Cameron (1930) points 940 M. J. SHUB out admirably the fallacious assumptions underlying most histo- logiwl methods of detecting calcium. It is obvious that the staining and micro-precipitation tests have not yet been developed to the point where they are capable of giv- ing reliable information regarding the amount and distribution of these constituents in microscopic sections. The interesting method of micro-incineration employed hy Polioard and his coworkers seems worthy of further application. Policard and Doubrow (1924) stained sections of tumors (presum- ably human) in the usual way and used adjacent sections for micro- incineration. They found that the more embryonal the type of amom tissue, judged by histological criteria, the less ash it ap- pears to have; also that degenerating and necrotic tumors give much more ash. They also observed that fibrous tissue is rich in calcium. Thus, although their results might be construed as fitting in with the commonly accepted notions regarding potassium and oaldum in cancer, the authors stated that the increased calcium may well be due to the increased amount of stroma fibrous tissue in such tumors. Ueing the same technic, Policard and Pillet (1925) stained see- tions of experimental tumors for calcium and compared its dietri- bution with the pattern of ash obtained by micro-incineration. They found that the zone of active, proliferating tissue at the pe- riphery of the tumors is very low in ash, especially in calcium, but that the necrotio zone at the center of the tumor is rich in mineral matter, especially in caloium. Soott and Horning (1982) applied this micro-incineration method to the study of hnman breast tumors, of mouee carcinoma 63, and of mouse sarcomas 37 and 180. In the case of the breast oarcinomas, “the infiltrating cancerous new growths are character- ized by a heavy deposition of mineral salts when viewed at low magnifloatione. Observation with oil immersion lenses shows that the nuclei contain rather more ash residue than do nuclei of the normal duct tissue. This deposit is concentrated along the pe- ripheral margins of the nuclei . . . and corresponds well with the hyperahromatization described by Horning and Richardson in ma- lignant growths. The nuclear inorganic salts contain visibly more iron oxide than do those of normal oells. “The remaining mineral deposits in the cytoplasm are more abundant than in the normal cells and contain an appreciable quantity of calcium and iron 8alts.” The “extraordinary concentration of the inorganic salts in the tumor tissue is well shown in the accompanying photomicrographs . . . , and was found to be a feature peculiar to both human and rodent neoplasms. . . . SODIUM, POTAWIUM, OILIAIUY AlVD MAGNESIUM IN OANOEB 941 ‘‘A survey of the inorganic struoture of neoplastic and normal tissues demonstrates conclusively that malignant growths are rioher in their mineral contents than normal tissues-especially in calcium and iron oxide.” m. BLOODANALYSE8 Bodhm, Potassium, Calcium, arad Magnesium Coderats : It was not until 1920 that accurate quantitative analyses of the individual inorganic constituents of blood in cancer were made. Prior to that time there were a few investigations of this phase of the subject, but they are of historical interest only, since the data were obtained by crude methods. Although Biernacki published his blood analyses in 1894, his results are still quoted, even by recent authors. He reported values for the sodium and potassium content of blood in a few cases of oarcinoma. These data are of little value, however, since they are for whole blood and not for serum or for plasma. Moore and Wilson (1906) ashed human blood serum and ti- trated it with standardized acid. They concluded from their ti- tration data that the “average alkalinity” of the inorganic salts is greater in cancerous than in healthy serum. No analyses for the individual inorganic constituents were made. A few years later Mottram (1910) examined the blood of cancer patienta for sodium and potassium. He determined the amount present by the length of time the alkali metal could be detected spectroscopically in the flame, and concluded that the blood of car- ainomatous persons contains more potassium than normal blood. Since, however, the method is only a crude one, and since the ex- aminations were made on whole blood, the data need not be cited here. Goldziieher (1912) and Goldzieher and Rosenthal (i913) ana- lyzed the blood of a few cancer patients for calcium. They used Voorhoeve’s (1911) method, which had an estimated accuracy of 10 per cent, and obtained values for CaO that ranged from 30 to 40 mg. per 100 C.C. for normal persons. (This corresponds to 21-29 mg. Ca per 100 C.C. ; the accepted normal value iff about 10 mg. Ca per 100 C.C. for serum or plasma and is less for whole blood.) The analyses were apparently made on whole blood. These authors re- ported that calcium values dightly lower than normal were ob- tained in cancer cases. Krehbiel (1920) used the method of Halverson and Bergeim (1917) in determining blood plasma calcium in 34 cases of ma- lignancy and in 43 non-cancerous patients. He concluded that in the cancer cases the average calcium values were within normal limits ;nothing characteristic of either the nature or of the location of the tumors was observed. Although the average value of plasma calcium in the malignant cases was 9.4 mg. per cent, the individual variations ranged from 5.9 to 15.1 mg. per cent. In the non-cancerous patients the values ranged from 4.3 to 14.3 mg. per oent. Since these were not cases of disturbed calcium or phos- phorus metabolism, the wide variations would indicate that the findings are inaccurate. The first critical investigation in this field was published by Waterman (1921), who believed that a study of the calcium, potas-

TABWII: InargMvta m~m~fbt8or Blood am (Wstennsn, 1081) Tlvmty-one oonrtltwnt Five NodPenom a*naer Patisntr

Potudom 10.7 mg. per 100 0.0. 81.3 mg. per 100 e.c. Wm 316 mg. par 100 e.0. 821 mg. per 100 a.c. woim 10.4 mg. per 100 0.e. 11.7 mg. per 100 ae. sium and sodium contents of blood serum was desirable in the caae of both tumor-bearing inhividuals and normal persons, especially since it was only recently that accurate microanalysis of these oonstituenta had become possible ;in his opinion, no value oould be attached to preceding communications on this subject because of the faulty methods employed. He showed the importance of avoid- ing even slight hemolysis if aorrect values are to be obtained for serum potassium : since the red cells are rich in potassium it is im- portant to remove them completely before analping serum. Waterman oentrifuged the serum twice to insure complete removal of the red cells. The individual values he obtained for blood aeruxn (human) .were given in full; the average values are shown in Table II. Waterman concluded that, contrary to many reports, the sodium, potassium, and calcium contents of the blood serum are not siflcantly changed in cancer. Aocording to Blum and Klotz (1923), the magnesium content of blood in cancer had not previously been studied. In fact, they stated that until the method of Hirth (1933) was pnblished, there was no means available for its determination. They determined plasma calcium and magnesium in duplicate, dimarding those re- sults which differed by more than 4 or 5 per cent. The normal values which they obtained by the methods of Hirth were 9.3 to 9.9 mg. per oent for calcium and 2.3 to 2.7 mg. per oent for magnedum. In 8 canwrous patients the calcium level was normal; in one it was slightly increased and in 2 it was slightly decreased. The mag- nesium values were normal in 7 cases of cancer and dightly in- creased in one cam. The work of Theis and Benedict (1924) constitutes an important contribution. These careful and accurate workers obtained the blood specimens from patients before breakfast, allowed the blood to dot, and separated the serum from the clot within two hours. The analyses for sodium, ’potassium, calcium, and magnesium were done on serum, for, as the authors stated, “Whole blood &res without aocompanying hematocrit readings are likely to lead to erroneous conclusions. ” Sodium, qotassium, and calcium were determined wording to Kramer and Tisdall ; the supernatant

Turn m:Zwrgoavb CMitwntr of Blood Em (Their and Benedict, 1984) Malignant Qrowth

Non-malignant Qrowth Early , Advanced t3nutituent Normal (0 WW) (la CMW) (14 cam) mg*per mg- psr mg. per mg. per 100 O.C. 100 C.C. 100 0.0. 100 Cd. Wum a86 aa7 aae a16 Pot.luium 20 81.4 20.1 20.0 aplafam ~.a-io.6 B.3 9.4 9.1 8.s 8.6 8.8 2.8 2.8 NaUI 670480 689 028 610

below below above Kind of Om 9.3 mg. per cent 8.2 mg. per cant 80 mu. per cent 081 mg. per cent Non-malignmt 50 per cent 38 per oent 16 per cant 60 per cent (0 -) Early Mrlignant 40 per cent 88 per cent 40 per cent 40 per oent (18 cam) Advanced MaIigpent 60 per cent 50 per cent 68 per cent 25 per oent (14 -1 eolution from the caldum determination was used for magnesium, which was analyzed amording to the method of Denis (1920). Chloride was determined according to the method of Friend (1922). The average values obtained with these methods for normal blood serum and for serum from cases of malignant and non-ma- lignant growths are given in Table IIT. The authors concluded that sodium and chloride are normal in cancer, that potassium is somewhat decreased in a large proportion of cases, and that the more advanced cases have lower average figures than the non-ma- Kgmnt cases; they also concluded that calcium ie low in cancer. 944 ILb. J. SHEds These conclusions were based on an analysis of the detailed data, which were published in full; their analysis is reproduced in Table N. In an attempt to correlate blood potassium and calcium with cancer, Renaud (1924) analyzed the blood of some 200 patients. He concluded that there are inconstant and irregular variations in the potassium and calcium contents of the blood of cancer patients at all stages of the disease. These irregularities and inconstancies are readily underfitandable since his determinations were made on whole blood. Beginning with 1925, reports on blood analyses in mcer be- came more numerous because of the widespread use of the recently devised micro-methods for serum potassium, calcium, sodium, and magnesium. However, not all of the published data are of equal value. Paolucci (1925) determined blood calcium by Cristol’s method in 24 cancer patients and in 7 normal persons. In each case esti- mations were also made by de Waard’s (1919) method which, Paolucci stated, oonsistently gave values about 20 per cent higher. The values he obtained for normals varied from 6.0 to 8.5 mg. per cent with an average of 7.2 mg. per cent. If the determinations were made on whole blood, the data are valueless ; if they were ob- tained on serum, they are incorrect. The data on cancerous blood need not, therefore, be cited. Leicher (1925) observed a slight lowering of serum calcium in 18 of 21 cases of carcinoma studied. For normals, he obtained values ranging between 10.9 and 12.0 mg. per cent; in the 18 cases of carcinoma the values ranged between 9.9 and 10.8 mg. per cent. Rohdenburg and Krehbiel (1922) analyzed whole blood of tumor-bearing rats for sodium, potassium, and calcium. ‘phey (1925) also removed various organs of internal secretion from rats and analyzed the blood for potassium, sodillII1, and calcium as be- fore, employing the methods of Kramer and Tisdall. The values were compared with the values obtained from normal control rats and from control rats having merely incised wounds. These de- terminations, too, were done on whole blood, though Theis and Benedict in 1924 had said: “Whole blood figures without accom- panying hematocrit readings are likely to lead to erroneous con- clusions.” Since no values were reported for blood cell volume, Rohdenburg and Krehbiel’s data cannot be used to calculate the concentration of these constituents in the plasma. Peculiar values for serum calcium were published by Svehla (1926). He reported that in canoer patients the serum dcium varied from 14.2 to 22.0 mg. per cent, and that in normal persons it ranged from 22.6 to 24.6 mg. per cent. Bince he used the reliable SODIUM, POTABIWM, OALOIUX AXVD MAGNESIUM IN OANOEB 946 Kramer-Tisdall method, these incredibly high values appear to be due to analytioal errors. Increased serum potassium values were obtained by Ottonello (1926), who also nsed the method of Kramer and Tisdall. Since, however, potassium values greater than normal were obtained in a number of non-cancerous conditions, Ottonello concluded that the estimation of serum potassium is of no diagnostic valne in cancer. The sera of 6 cancer patients were examined. The values varied from 40.5 to 85.2 mg. per cent, as compared with a range of 19.9 to 34.1 mg. per cent in 7 normal individuals. Rohdenburg, continuing his blood analyses in collaboration with Blumgsrten (1927), analyzed the blood of 173 patients for sodium, caloim, potassium, and magnesium. The method of Kramer and Gittelman (1924-25) was used for sodium, of Kramer and Tisdall for alcium and potassium, and of Bernhard and Beaver (1926) for magnesium. Seventeen cases of carcinoma were studied. Blum- garten and Rohdenburg concluded that “In carcinoma there was a strikingly low magnesium content and a low calcium content in 50 per cent of the cases. . . .” Unfortunately, in this investigation aa well as in the previous ones of Rohdenburg and Krehbiel (1922, 1925) the analyses were performed on whole blood; determinations of oell volume were not reported. In a careful piece of work, Jones and Rourke (1927) determined the sodium content of mouse plasma. At first they nsed the Kramer-Gittelman method, adapting it for the small quantity of plasma obtainable from a single mouse. The variations, however, “seemed too great to be real” and the results of 50 analyses were therefore discarded. Bourke (1928) then modified this method, and obtained results with small amounts of mouse plasma that were entirely satisfactory. With this modified teahnic, the plasma sodium in a large num- ber of mice was determined. Of the 101 mice analyzed, some strains were 100 per cent susceptible to a transplantable adeno- mroinoma of the mammary gland and some strains were 100 per wt immune. At least 10 mice were used in each group and indi- vidual analyses were obtained for each mouse. No significant dif- ferenoe was found in the plasma sodium of susceptible and non-ma- oeptible mice, or of tumor-bearing and non-tumor-bearing mice, or of non-eusceptible mice six days after inoculation with the tumor. The authors therefore concluded that, by itself, the sodium con- centration is not the determining factor in immunity or suscepti- bility, and pointed out that the increased sodium content reported by Itohdenburg and Krehbiel for whole blood may have been due merely to increased plasma volume and not to increased sodium content of plasma or cells. They suggested that it may be in its 946 X. J. 8- relation to other oatione that eodium is of importanoe, and etated: “Unfortunately, the small amount of plasma available from hdi- vidual mice does not permit acwurate analysis of calcium by any known method, so that we are at present unable to determine the importance of the eodiumoalaium ratio in respect to the growth of the inooulated tumore in mice.” A comparative etudy of potassium and calcium in oancer and in inflammation wae made by Fukunaga and Nagaoka (1928). They analyzed the blood eerum of 16 nodadult male rabbite and found that the calcium varied between 14.3 and 18.1, with an aver- age of 16.2 mg. per cent, and that the potaseium varied between 20.6 and 29.4, with an average of 25.8 mg. per cent. Potasaim waa determined aooording to the Kramer-Tisdall method and calcium amording to de Waard. Kato’s rabbit sarcoma was then im- planted into 6 of theee rabbite. Inflammation was produced by

(kleium Potaulom Pot.drrm: 0.lSiom Ratio Before After Before After Before After Indtion Inwdation Inwation Inoarrwion Inomlation Inoedation mg.per mg.psr mg.per mg.per Wbit 100 6.c. 100 C.O. 100 c.6. 300 c.c. Nlnnber 1 163 18.1 875 S4.l 1.71 1.84 ‘l 8 16.4 88.1 80.4 86.9 1.00 1.00 11 a 16.8 86.1 97.8 80.4 1.68 0.78 ‘l 4 17.9 19.1 80.6 aa.6 1.43 1.88 l1 6 15.7 88.1 86.1 88.7 1.66 1.08 ‘1 6 14.1 16.6 96.8 84.8 1.84 1.66 lycopodium in 5 rabbits and by kieselguhr in the remaiping 6 rab- bite. The blood eerum wae analyzed eubsequently at weekly in- t ervals. It wae found that in the tumor-bearing rabbits the eerum cal- cium inoreased and the potassium deareaeed. On the other hand, in the animals in whioh inflammation had been produoed, the po- taeeium remained at a conetant level; the calcium either did not change at all or showed only a alight transitory increase which subeided after two weeke. In them 10 rabbite the potaseium : cal- cium ratio was the eame at the end of four weeke I)B before the in- fiammation wae produced. T.he potassium : calcium ratio in the tumor-bearing rabbite decreaeed following inoculation, a8 ehown in Table V. Fukunaga and Nagnoka uited the work of Konoe (1924), who also ehowed that eerum potaaeium decreaees parallel with tumor (sarcoma) mowth. But Konoe noted neither an increaee nor a 8oDmM, POTASSIUM, 0- mi3 MAGNBSIUY IN UANUEB 947 deoreaee in serum calOium at any the. Fukunaga and Nagaoka stated that their calcium findings, therefore, were at variance with thme reported by Konoe. What experimental animal Konoe used was not stated. Booording to Morivek (1932), normal values were found for serum potassium by Konopleff (1928) in nbn-cacheotic patiente, and decreased values in oachectic patients. Goerner and Shafiroff (1928) analymd the serum of tumor- bearing rats for calcium moording to the Krmer-Tisdall method. In normal rats they found 11.0 mg. per cent for serum calcium, a value which agrees with that usually obtained. For tumor-bearing rat0 which had been treated with parathormone, the average of five determinations of serum calcium was 15.7 mg. per cent. No values were reported for the serum calcium of untreated tumor-bearing rats. Harnes (1929) inoculated the testicles of 30 rabbits with the malignant neoplasm described by Brown and Pearce (1923). The rabbits were divided into 2 groups: (a) 12 which died from the effects of the tumor; and (b) 18 whioh recovered. Blood analyses were made weakly for two months. In the rabbits that died he- cause of the tumor there was a decrease in the serum calcium, whereas in those that recovered there was no pronounced change in the dcium, 'Fhe literature prior to 1920 contains no reliable values for the inorganic constituents of blood serum in cancer, and, although papers on this subject have been appearing in increasing numbers Within the last few years, the present state of affairs is far from satisfactory. 8chepetinsky and Kafitin (1929) stated that the findings of numerous authors on the contents of calcium and po- tassium, obtained by various methods on various types of material, are so divergent that it is difficult to form an opinion regarding the changes in the mineral content of blood serum in cancer. Accord- ingly, they analyzed blood serum for calcium, potasdum, and sodium in a comprehensive series of cases which included 60 nor- mal women, 60 cases of fibroma, and 60 cases of carcinoma of the uterns. They determined calcium according to de Waard (1919), potassium according to Kramer and Tisdall, and sodium according to Miller (1909). Their observations were as followe : (a) Calcium in cases of uterine fibroids was practically the same a8 in normals; in a few cases it was slightly higher than nor- mal. In carcinoma cases the calcium was either normal or slightly below normal. (b) Potassium was usually normal in fibroid cases ; in about 33 per cent of the cases it wa8 somewhat below normal. In the ma- jority of the cancer cases the potassium was normal, but in 40 per cent of the cases it was somewhat below normal. 948 M. 3. SHBAB (c) Sodium fell witbin normal limits in both fibroid and car- cinoma cases. &hepetinsky and Kafitin oonoluded that the ~erum calcium in patients with tumors was normal, irrespective of whether the tumors were benign or malignant. Furthermore, no correlation was observed between the rate of growth of malignant tumors and serum calcium. Neither was any definite parallelism observed be- tween the rate of growth of the tumors and the lowered potassium content of the serum; the deoreased potassium is not speciflo for cancer since a similar proportion of low potassium values was found in the series of benign tumors. Finally, there was no in- crease in the ratio of potassium to calcium in malignanoy as far as the serum was concerned ;the same values were obtained as in cases of benign tumor.

TABU VI: Percentage DisCtlbutim or Turnor Cue8 Aooordhg to Missrd Content of tAa Bsruln (Bahepatinaky mnd Kotitia, 1929) Kind of Celdm~e Potaadnmmnge Tumor in mg. per 100 0.0. in mg. per 100 0.0.

Inspection of the data shows that the values for serum calcium in the benign tumor group ranged from 11.6 to 16.0 mg. per cent ;in 40 per cent of the cases it was greater than 14 mg. per cent. In the oancer group it ranged from 11.2 to 15.2 mg. per cent ;in 30 per cent of the cases it was greater than 14 mg. per oent. The authors state that these figures show that the calcium content of blood serum of women with fibroids and cancer is within the established range for normal women. However these values are sot normal values for serum calcium. Numerous papers have been published within the last decade giving a vast amount of data on the serum oaloium of normal men and women, and it is generally agreed that these values do not vary much from 10 mg. per cent. The calcium values of Schepetinsky and Kafltin are, therefore, too high. A summary of their data is shown in Table VI. Louros and Gaessler (1929) obtained a large amount of data on the blood of 15 women with carcinoma of the uterus. Among the many determinations were those for sodium (amording to Miiller), potassium and calcium (both according to Kramer and Tisdall), and magnesium (according to Gadient). They reported a large decrease in serum sodium as compared with normal, a large in- crease in serum potassium, normal serum calcium, and slightly elevated serum magnesium. 80DmMp POTABSIUX, OALUIWM AXD MAGNBSIUM IN OAXVOEB 949 This work was extended by Gaessler (1930), who performed similar determinations on the blood of 9 patients with carcinoma of the uterus before and after irradiation with x-rays and radium. From three to five spwimens of blood were analyzed in each case, the last analysis being made five months after irradiation in some instances. In 5 cases there was improvement after treatment, in 4 cases there was no improvement. The blood picture waa not strikingly different in the two groups either before or after irradia- tion. The calcium was about normal in 6 cases at the start ;in 3 cases, which did not respond to treatment, the calcium was curiously high at the start. Most of the oases gave very high calcium valueu im- .mediately after irradiation ; subsequently these fell toward nor- mal. The sodium and chloride values were all below normal, both before and after irradiation ;treatment produced no regular change in either. The potaesinm values were above normal in most cases prior to treatment ;no regular change was produced by irradiation. Beveral months after irradiation, lower values (normal or below) were obtained in most of the cases which showed improvement, whereas the 4 unimproved cases all gave high potassium values. Gaessler was of the opinion that the chemical picture is of im- portance in determining whether cure is effected. Even when clinioal disappearance of the tumor is noted, cure cannot be said to have been effected unless the blood chemistry has been restored to normal, he stated. In a careful piece of work De Fermo (1930) analyzed the blood serum of 46 cancer patients. The blood specimens were obtained before breakfast and the method of Kramer and Tisdall was em- ployed. In control cases the author obtained values for serum cal- cium which ranqed between 9.7 and 10.2, with an average of 9.9 mg. per cent. In the 30 cases of early cancer the calcium ranged be- tween 8.2 and 10.2; the average was 8.7. In the 16 advanced cases it varied between 4.0 and 7.0, with an average of 5.2 mg. per cent. Of the 30 early cases only 4 showed normal values; the average value in the other 26 was 8.4 mg. per cent. It is of interest to note that the 4 patients with normal values were all less than forty years old: all the other patients with early cancer were over forty. In the opinion of De Fermo a lowered serum calcium is not oharaateristic of the cancerous process itself. He cited other au- thors whose work wag said to indicate that the serum calcium in individuals more than forty years of age is about 1.5 mg. per cent less than in persons between twenty and forty years of age-a uo- mlled physiolofical hypocalceda. The definite lowering of serum calcium which De Femo observed in his early cases, therefore, he ascribed to the advanced age of the patients. The striking de- 950 M. J. SHBAB

orease observed in advanced oases he WCLBinclined to attribute to the accompanying cachexia. It is to be regretted that De Fermo did not publish in detail his data on non-cancerous individuals, in- cluding their ages. No &ange from the normal was noted by Roe and Dyer (1930) in the serum calcium of hens bearing Roue sarcoma. They deter- mined a number of the constituents of the blood of sarcomatous and normal hens. The procedure of Roe and Eahn (1929) was em- ployed in the caloium determinations; 4 analyses of normal sera gave the same average as 6 analyses of sera from tumor-bearing hens. In ((A Comparative 8tudy of Body Fluids in Canceroue and Non-Cancerous Individuals,” Pitts and Johnson (1 930) deter-. mined the calcium, potassinm, and sodium contents of blood serum

TABU VII: Anazpen’or B~OO~sstcm, ma BWW B~UM (pittr end.;rohnron, 1~80) Waiunl Potudrmr wa in mg. per 100 ae. in mg. per 100 C.C. in mg. per 100 C.G. Sawn Nodand non-cancerow fpcuviaaslr 11.0 (85). 19.5 (20) 337 (19) Urnaeroun individuals 11.3 (35) 21.2 (34) 339 (26) Blbtar FW Nod and non-cancerous individuals 9.7 (86) 16.B (14) 322 (6) ~eerowindividoalr 8.2 (26) 17.6 (12) 338 (8)

The numbern in parentheme give the number of detenninatiom. and of blister fluid in a fairly large number of cases. The analyti- cal methods employed were: Clark and Collip’a (1925) modifica- tion of the Kramer-Tisdall method for caloinm; Rourke’s (1928) modification of the Eramor-Gittelman method for sodium; and the Kramer-Tiadall method for potassium. In the calcium determina- tion the pH was adjusted to about 5.0, in accordance with the sug- gestion of Shohl (1922). The data obtained in this careful study are given in detail in 17 tables. The results are summarized in Table VII. The authors concluded that the blood serum of oancer patients contains essen- tially the same amounts of potassium, sodium, and caloium as non- cancerous serum, although the potassium values are perhaps a bit higher in the cancer cams, and that “Canoer is not more prevalent in individuals of any one blood type than in any other.” Increased potassium content of the serum was reported by R6mond and Cantegril(19300), who employed the method of Ham- burger (1915). In 14 normal individuals the potassinm values ranged from 18 to 40 mg. per cent, with an average of 29 mg. per cent. In 13 cancer patients an average of 33 mg. per cent was ob- tained. The authors concluded that there was a slight increase in potassium, but found no correlation between this increase and the clinical condition. This observation was investigated further in a study of the po- tassium content of the serum in 62 cancer cases. Values of 20 to 30 mg. per cent were obtained by Umond and Cantegril (19306) in 38 cases ;i.e. 61 per cent of the mses had what they considered a normal serum potassium. In some cases the value was slightly elevated. In only a few instances was it definitely above normal; in these caees the patients were in bad condition. The conclusion was drawn that variations in serum potassium are rare, and that when they do occur they appear to be due to causes other than the cancer process. The authors’ earlier conclusion that there wa8 a slight “hyperkali6mie” in canoerous serum was not substantiated in this larger series of 62 caees. In a continuation of their previous studies, R6mond and Cante- gril (1930~)carried out 48 determinations on the blood corpuscles. Yhey wet-ashed the material with a mixture of ifur uric and per- chloric wid8 and determined the potassium as in their earlier work. In 16 non-urine in oancer. He analyzed rat uMe both before and after inoculation with Flexner caroinoma. A total of 10 analyses prior BODIUIYC, POTASMUM, OALOIUX AND MAGNESIUM IN OANOER 963 to and 10 analyses after inoculation were made for each constitu- ent. The methods of Kramer and Tisdall were used for sodium, potassium, calcium, and magnesium. The urine contained more sodium and more potassium after inoculation, in most instances ; calcium and magnesium were unchanged four weeks after inocula- tion, but were increased somewhat several weeks later. Similar analyses were made of the urine of 3 normal persons and of 3 cancer patients. The sodium and the potassium in the urine were less in cancer than in normal urine, the calcium varied irregularly, and the magnesium appeared to be somewhat less than normal. lo&ed Cdcizlm: Stimulated by the findings of the earlier analysts who reported that actively growing tumors have a lower calcium content than slowly growing or receding tumors, a num- ber of investigators studied the calcium content of the blood in cancer as soon as reliable micro-methods became available. Con- flicting findings were reported, and although most of the analysts found decreased blood calcium values in cancer, the results were neither striking nor regular. The analytical methods all gave values for the total caloium con- tent of whole blood (or of serum or of plasma). It is generally considered that calcium is present in serum in several forms : (a) oaloinm ions ; (6) calcium proteinates ; and, possibly, (G) organic oslcium oompounds, similar to calcium citrate, that are un-ionized and difisible. When no clear-cut relation between total serum calcium and cancer was obtained, attention waR devoted to possible changes in the ionized fraction of the blood calcium. Thus, Waterman (1922) stated that even when the total amount of calcium is constant there may we11 be variations in the amountR of these fractions. He discussed the three methods currently used to determine the calcium ion concentration of serum. The ohjec- tions to the methods of compensation dialysis and of ultrafiltration led him to conclude that there remains onlv an indirect method of estimating this quantity. He employed the formula of Rona and Takahashi (1913) in which [H'l [Ca++l = 350 X [HCO,'] ' where [Ca++] = calcium ion concentration [H+] -hydrogen ion concentration [HCO,'] =bicarbonate ion concentration Waterman determined the CO, content of the blood plasma of 49 cancer patients and obtained an average of 72 volumes per cent 964 Me J. SHZAB CO, as mrnpared with 65 rolurnee per cent for normal individuals. Although the values in the two groups overlapped, there was a 10 per cent difference between the averages, and Waterman concluded that there was a slight but definite shift towards the alkaline side in cancerous serum. He determined plasma pH in 27 cancer cases and found a shift towards a higher pH as compared with values obtained in nodpersons. Utilizing the above formula, Waterman conoluded that a 10 per cent increase in the ooncentration of biaarbonate in cancer would entail a corresponding decrease in the calcium ion concentration. Aacording to the formula, an increase in pH would also entail a decrease in the calcium ion concentration. R6mond, Lsendrail and Lasaalle (1928) concluded that the cal- cium ion concentration of plasma homes lowered when cancer ap- pears and riees again later, espeOially when the growth nndergoes regression. They studied tar canmr in rabbits and gave figures for the calcium ion concentration of the sgrum of the tumor-bearing animals. They pointed out that their work conflrmed the thesis proposed by Policard and others, and conoluded that calcium ion exerts an inhibiting effect on the cancerous procese. Tt should be noted, however, that them authors did not determine the calcium ion conoentration. Like Waterman, they determined plasma pH and CO, and obtained the calcium ion concentration by caleulation from the formula of Rona and Takahashi. The values which they gave for the calcium ion concentration ranged all the way from 21.5 to 98, presumably in milligram8 per liter. Boffo and Correa (1925), starting from the widely held View that there is a defloiency of ealcium in cancer, determined the amount of ultrafilterable caloium in the plasma of 16 omcer pa- tients. They found that about 62 per oent of the total calcium was ultrdterable, that the values for ultrafilterable calcium varied but little among themselves, and that they were not muoh different from those fokd for normal plasma. McDonald (1927,1929), too, called attention to the significance of oalcium ion in cancer. He suggested that cancer is due to a preponderance of univalent cations, auch as sodium and potassium, over bivalent cations, suoh as caloium and magnesium. He also suggested that radiation exerts its beneficial action in the treat- ment of cancer “by stimulating the ionization of the bivalent Rub- stances (particularly cdcium.) ” Hie article contains no data on ionized caloium. Reding (1927) has published a large amount of data on the oalcium ion concentration of blood plasma in oancer. In an ex- tensive investigation, he reported values for plasma pH, CO, and Caw in a large series of patients. The average caloium ion con- SODIUM, POTABSIUX, OALOIUM AND YAQNESIUM IN OANOEB 966 oentration obtained in the control group, which oonsisted of 25 persons, waa 22.4 mg. per liter, whereas the average in the 1111- treated cancer group (41 cases) was 16.8 mg. per liter. Detailed data are given for groups of patients before and after various types of treatment, i.e., operation, x-ray therapy, and radium therapy. Analogous fignres are given for those cured, those im- proved, those unimproved, and those who suffered relapaes. In a subsequent paper Beding and Slosse (1929) stated that their oonclusions were based on about 600 observations of pH, CO, and Cat+. Inspection of their detailed data creates a favorable impression ; the results look convincing. Nevertheless, their fig- urea for calcium ion ooncentration are of doubtful value. In com- mon with most other investigators they used the formula of Rona and Takahashi. As will be shown later, this formula is not ap- plioable to blood serum or plasma (see page 1004). The work of Beregoff (1930) is open to similar criticism. She studied a series of 281 cancer patients and compared these with a series of 25 individuals not suffering from cancer. Like Reding, she found a decreased calcium ion concentration in the blood l of cancer patients ; the average for the 281 cancer cases was 17.5 mg. per liter, as compared with an average of 22.6 mg. per liter for the normal controls. Like Reding, Beregoff calculated the calcium ion values from the formula of Rona and Takahashi. kther and Greenberg (1930~)also investigated the question of ionio oalcium in cancer. They determined the total and dif- fusible caloium of the serum of 15 patients with cancer. “By the diffusible calcium is meant that portion of the serum calcium which will diffuse through a collodion membrane completely permeable to crystalloids. Experimental work ha8 shown that the diffusible caloium fraction of the blood serum is not entirely ionized, but that it does contain essentially the same concentration of ionic calcinm a~ in the blood.’’ Calcium was determined by Tisdall’s (1923) modiffoation of the Kramer-Tisdall method. Inorganic phos- phone was determined by the Fiske-Subbarow (1925) method. It was found that, although 7 of the patients gave defhitely low values for total serum calcium, the values for diffusible calcium in the 15 patienjs were within normal limits. hnther and Greenberg also determined the serum proteins. They say: “The fact that the fall of the serum proteins is con- oodtant with a qualitative fall in the nondiffuaible calcium offers an explanation for the apparent deficiency of the blood calcium as seen in the total serum calcium values.” They concluded that blood calcium is apparently not a factor in the etiology of malig- nant diseases. 1No rtatement WM found III ta whether the detdnationa of pH and of 00, and the calaulationr of C.++were made on plawa, on warn, or on whole blood. 9S6 M. J. SHMB In a previous paper Ounther and Gheenberg (193Ob)had found that the normal concentration of diffusible calcium of the blood serum usually varied between 4.5 and 6 mg. per 100 c.c., with an average of 5.0 mg. per 100 C.C. ; the extreme limits of variation were 4.2 and 6.8 mg. per 100 C.C. The values obtained for diffusible cal- cium in cancer were fairly constant and were about normal. Therefore, the low values of total calcium were due to decreases in the non-difisible fraction, as these authors stated. However, a parallel decrease in protein and non-diffusible calcium in cancer was not definitely established. In several instances the fall of the serum proteins was not ooncomitant with a fall in the non-difisible calcium. Thus 3 case^ with the same protein content of 6.6,6.7 and 6.6 per cent gave values of 4.8,4.1 and 2.8 mg. per cent respectively for non-diffusi- ble calcium. On the other hand, a patient with a protein of 5.6 per cent gave a non-diffusible calcium of 5.3 mg. per cent. TABLEVm: Difuaibls C&um iw Carasrwr 8wm (Quntber and Qrmnborg, 1980) Total Mff dble Non-INYdble Total oalaiam WUm oslasom Rotain 0.8 mg. per 100 0.0. 4.5 mg. per 100 c.6. 4.8 mg. per 100 LO. 6.6 per Gent 8.6 4.5 4.1 6.7 7.6 4.8 2.8 6.6 10.5 5.2 5.3 6.6 7.7 5.1 8.6 6.8 8.1 4.0 4.1 6.4

Table VIII summarizes the data for 6 of the 12 cases for which protein analyses were reported. It is seen that large decreases in total calcium were found in cases in which no fall in protein had occurred; in these cases a pronounced decrease in non-diffusible caloium was noted unaccompanied by a concomitant fall in protein. Conversely, a comparatively low serum protein was noted in the case with the comparatively high total calcium.

IV. EFFP~CTOF SODIUM, POTASSIUM AND CALUIUM The effect of inorganic substances in cancer h.as been studied from numerous angles, in man and in experimental animals. These chemicals have been administered both orally and by injec- tion. By the oral route they have been given alone in the form of solutions or have been incorporated in the diet. They have been injected intraperitoneally, intramuscularly, subcutaneously, and intravenously. Attempts have been made to affect susceptibility and to alter the rate of growth of the tumor. Finally, tumors have been treated i~ vitro with various solutions prior to transplanta- tion in an attempt to change the usual course of evente. SODIUM, POTASSIUM, (IALOITJIY AXD MAGNMSIUM IN OANOEB 957 Clowes and Frisbie (1905) performed a few experiments on the effect of isotonic solutions of potassium and of calcium upon the development of tumors and observed that “mice inoculated and fed with potassium-holding materials appear to be more SUB- ceptible than those inoculated and fed with calcium material.” NBgre (1910)dipped bread into 0.6 per cent solutions of various dts, including sodium chloride, potassium chloride, and calcium chloride, and fed it to mice for four months previous to inoculation with adenocarcinoma ‘‘B.” The tumors that took were subse- quently inoculated into other groups of mice which had been given them various salts, and also transplanted into control mice. In general, a tumor from a mouse fed on a given salt was found to take in a greater percentage of cases in mice fed on that salt than in mice fed on other salts. In his conclusions NBgre stated that whereas this tumor was successfully transplanted in normal mice in 90 to 100 per cent of the cases, less than 50 per cent takes were obtained on inoculation into mice that had been fed for three or

Group Number of Yioe Average Weight of Tumor PMum treated ...... 15 0.94 gm. klrrillratrerted ...... 15 0.59 gm. aamtrolr ...... 15 0.10gm. . four months on a different salt. He also concluded that sodium ohloride has an unfavorable effect on tumor growth. More than a dmade later Nagre (1922)called attention to this early work of his and pointed out that he had shown that addition of potacrsium salts to the diet resulted in a larger number of takes and in an increase in the size of the tumors as compared with con- trols, while sodium ohloride had an unfavorable effect. This en- tire investigation, however, is of questionable value since Nagre need only 4 and 5 mice in each group, and since the results were not ulear-uut. Qsld5ieher (1912) injwted a 5 per cent solution of potassium dtrate subantaneonsly into mice bearing carcinomas (Ehrlich and Bsshford) and compared the results with those obtained by similar treatment with a 5 per cent solution of calcium lactate and with the results in untreated controls. The calcium salt did not affect the number of takes but it reduced the weight of the tumors as uompared with the controls, whereas the potassium salt increased the weight and size of the tumor& The decrease in size obtained with ualcium was not spectmular, but was consistently regular. The experiments were done repeatedly, always with the same re- 968 M. J. 6- aulte. One hundred and fifty mioe were used. Inoreasing the number of mice to 200, Golddeher and Rosenthal (1913) arrived at the same oonclusions, namely that cdoium hiders and potas- sium stimulates tumor growth. In one series of experiments the results shown in Table IX were obtained. These investigators also studied the relation between endo- crines and tumor growth; they used castrated doe and treated other groups of mice with various gland extracts. All reeulte were negative except those obtained with parathyroid extraot, whioh gave results similar to thoee obtained with injection of calcium salts. Ooldzieh’er and Rosenthal dted the work of Lewh and Anderson, who found that suboutaneons injection of potsemum hexatantalate resulted in astonishingly rapid growth of a rat sarooma. They failed, however; to state where this work was published. Robinson (1917) raised the question as to whether variations in the oonoentration of sodium ohloride in blood serum may be the cause of cancer. In a later paper he (1918) reoommended the use of potassium nitrate in the treatment of oanoer. In neither of these papers did he report any experiments or present any data. In a third paper he (1920) gave 8 case reports in whioh good re- sults were claimed. These patients had been treated with injec- tions of potassium nitrate supplemented “with mineral salts suoh as magnesb, oaloium, manganese, eta, with the object of replaa- ing the cell loss of these substanoes”! Sugiura and Benedict (1922), testing Robinson’s suggestions, obtained no effeot with either sodium ohloride or potassium nitrate. The work of Cramer (1918) is of oonsiderable hnportanae. This careful and thorough investigator treated pieoes of tumor tissue in vitro with solutions of eodium ohloride and of d&um ohloride and then inmulated them into doe. He found that the oaloium salt produoed a dietinot inhibition in growth, that treat- ment with the sodium salt resulted in no obvious ohange, and that the inhibitory &ion of the oaloium eould be oounteraoted by SOdiUm. Cramer used mouse carcinoma 63 of the Imperial Canaer Re- searoh hnd whioh, he stated, was a rapidly growing tumor that gave 90 to 100 per eent takes. He prepared an emulsion of the tumor and immersed part in M/7.6 NaC1, part in Mn.5 CaCl,, and kept an untreated portion as a oontrol.. The immersion time varied from 70 to 110 minutes. These emulsions were then in- ooulated into mice. “The experiments gave a uniform result in showing a dietinot inhibition of the growth of those oelle whioh had been treated with oalcium chloride. This inhibition manifests itself both in a diminution of the number of positive inooulationa 8OD#&, POTWM, UALdmM AND MAGNBSIUM IN OANOEB 068 and also in a lessened rate of growth of those tumours which had taken.” The cells treated with sodium chloride behaved like the control oeb. Cells treated with the calcium chloride solution and subse- quently treated for an equal length of time with the sodium ohloride solution “took again in 100% of cases and showed a rate of growth almost, although not quite, as high as that of the cells subjeoted to the influence of the Na ions alone. This removal of the inhibition induced by Ca ions is all the more striking when one takes into consideration that the experimental manipulations and more especially the length of time during which the cells are being kept outeide the body, are in themselves conditions not favourable to their subsequent growth. These experiments estab- lieh therefore clearly the existence of an antagonism between Na and Ca ions with reference to growth.” The inhibition produced by calcium is transient. Two weeks after implantation the tumors resume the rapid growth char- mteristic of strain 63. “Similarly if the tumours of the Ca strain be reinoculated after two weeks, without further experimental in- terference, into fresh mice, these inoculations take again in 100% of cases and grow as well as, if not better than, inoculations made from the control batoh or from the Na strains.” This experiment was continued throhgh 5 generations, and the tumor cells were treated with calcium chloride before inoculation in each generation. No permanent effect was obtained, although the same transient inhibition was noted after each treatment. “Histologically” Cramer found ‘(the protoplasm of the cells treated with CaC1, solution appears denser and less spongy than that of the cells treated with NaCl solution” or than the proto- plasm of the untreated cells. The number of animals employed in this investigation is not stated, but, from the illustrative ex- periments cited, a sdlicient number appear to have been used. Katase, Kawamura and Mizutani (1920) found that the intra- peritoneal injection of calaium solutions prevented the sumessful taking of tramplanted carainomas in mice. This protective effect waa attributed to a lymphocytosis produced by the doium. The peroentage of takes of a rat sarcoma was reduced following ad- mini~trationof aalcium. This summary of the findinga of these Japanese investigators is cited from Hiindel (1924), since their originsl article was not available. Katase (1922) subsequently stated that the growth-inhibiting effect of calcium and of sodim salts is due to the production of a lymphocytosis and that the growth-stimulating effect of magnesium and of potassium salts is due to lymphopenia. No data were reported in support of these statements. 4s In the earlier work on the effect of immersion of tumor tieeue in salt solutions little or no attention waa paid to the pH of the solutions employed. This important factor was taken into ac- count in the oareful study of Sugiura, Noyes and Fak (1921). They used phosphate buffersr, varying the pH, and varying the total phosphate ooncentration from 0.6 to 1.1. per cent. Tumor tissue (F’lexner-Jobling rat csroinoma) waa immersed in solutions of various pH and oontaining added salts for periods ranging from one to seventy-two hours. The effeot was then tested by in- ooulation of the tissues into rats. Immersion in phosphate solution at pH 7.0 for twenty-four hours was without effect; the tumors grew normally. After im- mersion in phosphate solutions of pE 6.8 and 5.1 growth was in- hibited. On the alkaline side, immersion in phosphate solution with a pH of 8.2 resulted in partial inhibition; immersion in a similar solution with a pH of 8.8 resulted in total inhibition. Similar treatment with sodium cbloride (0.15 M or 0.85 per cent) at pH 7.0 for twenty-four hours had no effwt; seventy-two hours immersion produced complete inhibition. Immersion in 0.078 M oaloium chloride at pH 7.0 for half an hour had no effeot ; after immersion for five and ten hours partial inhibition resulted, and eimilar treatment for twenty-four hours produced complete inhi bition. More dilute solutions of calcium chloride were also used at pH 7.0. Treatment with 0.009 M calcium chloride for twenty-four hours gave complete inhibition, as did immersion in 0.003 M cal- oium chloride for seventy-two hours. On the other hand, immer- sion for seventy-two hours in a Rifiger-Locke solutior which CON- thed the 8ame amomt of cdcizcm chloride (0.00S M) did sot affect the growth. This solution contained 0.16 M sodium ohlor- ide, 0.003 M potassium chloride, and 0.003 M calcium chloride ;the pH was 7.0. These exceedingly interesting findings, together with those of Cramer (5918),are worthy of further investigation. Solutions of inorganic salts were used by Rorrel, de Conlon and BoeE (1922) in an “ionoth6rapie Blectrique.” They attempted to influence the growth of a transplantable rat sarcoma which gave 100 per cent takes, and which did not undergo spontaneous regres- sion. The solution was applied to a piece of cotton which was placed on the tumor and covered with the positive electrode; the negative electrode was applied to the opposite side of the animal’s body. This method of administration was employed with solu- tions of 10 inorganic salts. A total of 40 rats were used, about 3 for each solution. The authora reported 13 66cures,y’none of which, however, was given by the calcium, potassium, or sodium solution. SODIUM, POTASSIUM, OALOXUM AND MAGNESIUM IN OANOEB hussy and Wolf (1922) immersed sections of mouse tumors in salt solutions and studied the cytology of the treated material and the effect of this treatment on the behavior of grafts. They worked with a mammary epithelioma, an atypical glandular epi- thelioma, and a sarcoma, all of mice. No mention is made of pH adjnetment of the solutions employed. With 0.7 per cent potas- sium ohloride solution the immersion period was varied up to four days, The cell protoplasm seemed to have been diluted; the cell volume was considerably greater and the outlines were irregular. However, mitoses were noticed but rarely. Ameboid motion was almost as pronounced as in normal cells. Inoculation of grafts gave positive results in three days ; the developmental period was muoh reduced, to as little as three days, without affecting the later development. The number of mice used is not stated. With 0.8 per cent calcium chloride, immersion caused the cells to home much smaller and rounder; the protoplasm was con- trmted, and ameboid motion was completely halted. Inoculation gave positive results in a few days, but a very noticeable increase in the length of .the developmental period was obtained. This period lasted a8 long as three weeks and resorption of the graft finally resulted. The number of animals used is not stated: the results are given in percentages. Troisier and Wolf (1922) worked with an adenocarcinoma of the breast which had appeared spontaneously in their mouse oolony. It was an active tumor which gave about 90 per cent suc- oemful takes on transplanting. After passing this tumor for 3 generations they tested the effect of immersion in solutions of po- tassium chloride and of calcium chloride. They used a 0.7 per cent solution of potassium chloride and a 0.48 per cent solution of cal- cium chloride. Treatment with the calcium solution delayed the appearance of the tumor for from five to ten days, depending on the duration of the immersion. It also decreased the number of positive takes. Onae the tumors appeared, however, they grew as rapidly as un- treated tumors; in four weeks they reached the same size as the oontrols. The subsequent development was not modifled. Immersion in the potassium solution resulted in a shortening of the latent period; the tumors appeared between the fourth and seventh days after inoculation as compared with an incubation period of about ten days for the control tumor tissue. The potas- sium-treated tumors attained the maximum size towards the fourth week. Their subsequent development was not modified. Thus these two salts exerted their effect solely on the first phase of development, the authors state. In this study, 19 mice were inoculated with calcium chloride- 962 M. J. BHElbB treated tissue, 19 with potassium chloride-treated tissue, and 11 with untreated tissue. The time of immersion was stated to be “plus ou moins prolongik.’’ Troisier and Wolf made no mention of adjusting the pH of their solutions. The next year Wolf (1923) published further immersion ex- periments : he again stated that caloium slowed down growth while potassium acoelerated it. He found that, while normal tissues have a fairly constant capacity for absorption, cancer tissue has a markedly exaggerated capacity, being able to absorb 3 times the amount of potassium and 20 times the amount of calcium originally present. Cytological study showed that potassium produced a “dilution” and calcium a “condensation” of the cytoplasm. Wolf concluded that calcium is absorbed to a remarkable extent by tumor tissue, that it exerts an inhibiting effect, and that it ap- pears to be capable of acting effectively only in the form of calcium ions. He recommended the use of ionic calcium in therapy, in con- junction with radiation. No details of these experiments are given by Wolf. The nm- ber of tumors analyzed before and after immersion is not stated and no data as to the calcium and potassium contents of the tumor tissue in the absorption experiments are given. In addition to their other experiments dealing with the rale of salts in cancer, Rohdenburg and Krehbiel (1922) found that “In- jections of the various soluble Ralts of potassium, sodium, and cal- cium, also failed to cause any recession of growth in mice bearing Crocker Fund sarcoma 180. Other mice repeatedly injwted before inooulation with these same salts and then inooulated at the in- jected area, failed to show the development of any resistance to the growth of this same tumor strain.” No details are given. Hoshino (1922) found that calcium chloride interrupted the growth of transplanted rat sarcoma, whereas magnesium chloride enhanced the growth. A mixture of these mbstanaes prevented growth of the tumor, Since Hoshino’s paper was available only in abstract, no details can be given here. The most elaborate investigation of the effect of oral adminis- tration of sodium, potassium, and calcium salts is that of Sugiura and Benediot (1922). They state: “Since it is possible to affect definitely the virulence of tumor cells by chemical treatment out- side the body , . . we have felt that extended studies along the line of possible chemotherapy in cancer are imperative. For this pur- pose prolonged oral feeding of inorganic salts has many advan- tages over intravenous injection.” The baddiet which they used consisted of 32 parts of cracker meal and 5 parts of whole milk powder. Preliminary experiments showed this “to be an ideal diet for the growth, maintenance, re- WDIUM, WTMEIIUM, CAWlIUM AND MAGNEBIUM IN OANOEB 963 production, and perfect milk production of the albino rats. The completeness of the diet was further proved by the normal rate of growth of the transplanted Flexner-Jobling rat carcinoma in rats fed on this diet, controls being fed upon a common diet of bread, milk, and carrots.yy Throughout the experiments the special diets (basal diet + spe- cial salts) were fed for from seven to fourteen days prior to in- oculation, “so that the administered salts might have their full physiological effect both before and after inoculation with the tu- mor tissue. This procedure also served to permit the experimental animals to become accustomed to the new diet. As a control, the same number of animals of nearly equal weight were fed with our basal diet and were inoculated with the same tumor tissue at the same time. ’’ The authors studied the effect of 32 salts upon the Flexner- Jobling rat carcinoma. The effect of these salts on the growth, weight, and general condition of normal, non-cancerous rats was first investigated. These salts were then fed to rats and, after the stated period, the rats were inoculated with the tumor. Several levels of concentration of the salts were tried. Potassium carbonate and calcium chloride both showed a re- tarding influence on the growth of the tumor, although the effect was not striking. Since in all these animals the general condition was very poor, it might be concluded, the authors state, that they were apparently “dealing with a non-specific effect upon the tu- mor-an effect which is simply due to the general poor condition of the host. Our results with tellurium feeding lead us to question the correctness of such a view. Tellurium causes a most marked effect upon the general condition of the animal. Growth ceases, the hair falls out, eye infections develop, etc. Yet these emaciated, denuded, undersized animals will show just as good a growth of cancer tissue a8 do those in a state of perfect health. . . . Such a result seems to indicate that there actually are some fundamentally different conditions governing the maintenance and growth of so- matic and tumor cells.” The conclusion is reached that “calcium chloride probably has a slight selective action upon tumor ‘growth.” No effect was obtained with either potassium nitrate or sodium chloride upon the development of grafted tumors. The authors state: “Our experimental results do not verify Robins’ [Robin- son’s] hdings, as potassium nitrate had neither immuddng influ- ence in rats nor a retarding influence upon the growth of the trans- planted cancer. It is also to be noted that this salt had no inhibi- tory influence on the appearance of secondary tumors.” Waterman (1923) treated normal and malignant tissues of mice with salt solutions, using an electrical method of application. The 964 Me J. 0- solutions were applied to the skin, and electrodes were employed in an attempt to cause the salts to enter the tissues. A current of 10 milliamperes was .used, with a potential of 40 to 60, volts. Uni- valent cations showed little or no effect when administered in this way. Bivalent cations had a favorable effect on tar mcers, dif- fering only in degree, in the following order : magnesium, calcium, hc,cadmium, and lead. The effect of the heavy metals was much greater than that of calcium or magnesium, The work of Mend616eff (1924) is sometimes cited as contra- dicting the widespread opinion regarding the inhibitory effect of calcium on the growth of malignant tissues. In thie work, solu- tions of potassium peptonate and calcium peptonate were injected into the veins of guinea-pigs, and the sera obtained from these animals were used as culture media for embryo skin. Mend616eff concluded that these experiments, which were reproduoible to a striking degree, showed that caloium iona in elevated conaentra- tions appear to be the true stimulating agents of cell growth in the case of embryonic tissue. Since, however, no results for tumor tissue were reported in this paper, it cannot be cited as showing that caloium has a stimulating effect on tumor growth. In studying the Gundermann-Duttmann test for gastric ulcer whioh is based on the relationship between the intake and excre- tion of water and sodium chloride, Schlesinger (1923) found that the sodium chloride exoretion is normally so variable that this test cannot be used for the diagnosis of mcer. Sodium ohloride re- tention omurs in carcinoma of the stomaah and in malignant tumors of other organs, but it also occurs in pyloric stenosis and, therefore, is not pathognomonic for weer, The effect of feeding potassium and calcium saltg was studied by Hiindel (1924). To a basal salt-poor diet he added various salts and fed the resulting mixtures to groups of mice for two weeks. He then inoculated the mice with a transplantable car- cinoma. Group I (25 mice) received an adequate mixture of salts in addition to the basal diet; this served as the control group. Group II (34 mice) received a supplement of sodium ohloride. In Group 111 (36 mice) the supplement waa a mixture of potassium ohloride and sodium chloride. In Group IV the supplement was a mixture of calcium chloride and sodium chloride. In the control group the tumor took in 32 per cent of the mim; in the sodium chloride group there were 29 per cent takes; in the potassium chloride group 39 per cent, and in the calcium chloride group 27 per cent. Hhdel concluded that the feeding of potassium in- creases the number of takes and promotes growth, wheress feed- ing of calcium causes a reduction in the number of takes and a diminntion in the rate of growth. These conclusions are not con- SODIUM, POTASBIUM, OALOIUM'AND MAGNESIUM IN OANoBII1 966 vinoing, however, because the differences between the percentages of successful takes in the various,groups were slight. In contradistinction to the fhdinge of previous investigators, Kimura and Wada (1924) reported that treatment of tumors with calcium in vitro was without effect. They immersed pieces of Jensen rat sarcoma, Flexner carcinoma, Kato rabbit sarcoma, and Kyoto fowl sarcoma in solutions of calcium chloride and of other salts. The effects of the different ions were compared by using solutions approximately isotonic with 0.85 per cent sodium chlo- ride. The effect of varying the concentration of sodium ohloride and of varying the pH waa also studied. Tumor tissue treated in this way was then inoculated into animals and the growth energy observed. The authors found that the rate of growth of the tumors was not retarded by calcium or sodium to any great extent. The number of animals used is not stated. Renaud (1924)treated patients with calcium gluconate but, in spite of repeated injections extending over a long time, all the cams went to a fatal termination; the calcium salt did not cause recession or prevent metastases. The number of patients treated is not stated. Treatment of cancer patients with calcium chloride was found to produce unfavorable results by 810SS0 and Reding (1924). They treated with salt solutions 40 patients who had been given up by surgeons and x-ray men as hopeless. The salts were em- ployed alone, in mixtures, and with adrenalin. Slosse and Reding claimed to have obtained good results when the salts were given together, and reported 16 cases. They state that the action of the various metal ions was evident in 50 per cent of the cases and that they considered calcium to be the only one of the bivalent cations which exerted an unfavorable effect. They injected calcium chlo- ride intravenously in 10 patients with cancer. Of these, only one improved (for a short period of twenty days), 3 died in a few weeks, and in the other cases the action of calcium, although less pronounced, was also harmful. Reding and Slosse subsequently came to a different conclusion as regards the action of calcium. Reding (1928) reported a large amount of data purporting to show that there is a decrease in the oaloium ion concentration of the serum in cancer patients. Red- ing and Slosse (1929), therefore, treated patients with parathor- mone in order to increase the calcium ion concentration. They ooncluded that certain patients have a predisposition to cancer, that this predisposition is hereditary, and that it is caused by an endocrine deficiency. They took the position that parathyroid medication may be employed as a logical preventive measure for precancerous conditions, in relatives of cancer patients, and after operation. 986 116. J. BHEAB The effeot of cations on tieeue growth im uitro waa etudied by Boffo (192&25b), who cdtivated nodtiaene (aok embryo heart) and neoplastio tissue (rat earooma). Chioken and rat plasma were employed ae oulture media. Solutions of potassium ohloride were added in varioue amounts to the plasma; eimilar experimenta were performed with mlcinm chloride and with mag- nesium ohloride. The experiments were performed repeatedly and oomparable reeulta were obtained regularly. The reeults are snmmnrized in tablee. Two groupe of experiments were carried out. In the ftret group a oomparieon was made of the effect8 of eolutions containing equal percentages of theee three salte. In the eeoond group the effeote of equimolar eolutione were compared. In both sets of experiment8 similar reeults were obtained: good growth waa ob- served when varioue concentrations of potaaeium ohloride were ueed; magneeium ohloride inhibited growth exoept in the more dilute solutione; caloium chloride had an even greater inhibiting effeot. The reeulte obtained with normal and neoplaatio tiesue were qualitatively alike ae regard8 the relative inhibiting effeot of the three cations. An important quantitative difference was noted, however: leee dcium chloride wm required to inbibit the growth of tumor tiesue than of nodtieene. Roffo (192P25c) followed thie work by an intereating investi- gation in whioh ealta were added to plasma in uiuo. He injected solutione of potaesium ohloride, of calcium ohlor’ide, and of mag- neeium ohloride into RnimRle and ueed the plasma aa a culture medium. As a oontrol medium he used normal plasma to whiah Ringer’e eolution waa added. In the &st experiment chick embryo heart wae cultured in eaoh oaae in quadrplimte. It waa found that plaema from ohidrene treated with potassium chloride gave aa good growth ae control plaama plue Binger’e eolutien. Plasma from ohickene treated with oaloium ohloride gave no growth. With plasma from chiok- en8 which had received magneeium chloride, diminished growth waa obtained. In a eeaond experiment a eimilar produre waa followed in culturing rat earcoma. Plaema from animale treated with po- taeeium ohloride gave enhand growth; plasma from animale reoeiving magnesium ohloride gave reduced growth; and that from oalcium chloride-treated animals little or no growth. All epeCi- mens were compared with tiieue grown in nodoontrol plaama to which Binger’e solution had been added. Culturee were run in quadruplicate and.3 implants were made in each of the quadrupli- cate solutions. In a third experiment potaaeium peptonate wm injeoted into SODIUM, POTAWIUX, adtaIUX AND MAGNEBIUM IN UANOEB 967 animals and the plasma was used as a culture medium; the same was done with calcium peptonate. Here, too, plasma from PO- tassium-treated animals enhanced the growth of normal and cancer tieaue as compared with control plasma plus Ringer; and again serum from calcium-treated animals gave either little or no growth. In the previous experiments calcium chloride and potassium ohloride were injected intravenously. When these solutions were injected subcutaneously and the plasma used for culturing normal and malignant tissue, results were obtained that were similar to those obtained on intravenous injection. Similar results were also obtained when plasma of normal chickens was mixed with plasma from salt-injected rats instead of plasma from normal rats mixed with that from salt-injected chickens. Roffo and Encina (192P25) immersed portions of rat tumors in 13erum ifi vitro and then inoculated them into rats. They used serum from normal and cancerous men, from’horses, and from normal and tumor-bearing rats. In general, they found that after immeraion in serum from normal animals the tumors did not take, whereas after treatment with serum from a tumor-bearing animal, the tumor took successfully. The experiments were performed repeatedly, using up to 12 rats for each -specimen of serum tested. After immersion in canoerous serum for twenty-four hours, the tumors almost invariably took ; after immersion for forty-eight hours failures were more frequent. In still another experiment, a specimen of normal human ‘serum was divided into 3 portions: one was the control; to the second patassium chloride was added to give a concentration of 0.018 per oent ;and to the third portion calcium chloride, to.give a concentra- tion of 0.012 per cent. The tumor fragments immersed for twenty-four hours in the untreated and in the calcium chloride- serum failed to take, on inoculation into rats, while the potassium ohloride-serum gave successful takes. Repetition of the experi- ment with another specimen of normal serum gave similar results. This phenomenon was further investigated by Roffo (1925) in ouIture experiments with rat carcinoma and rat sarcoma as well a13with chick embryo heart. Chicken serum and Ringer’s solution were used as culture media. Various amounts of potassium or oaloium, or of both, were added to the serum; and varying amounts of potassium chloride (0 to 0.58 g. per liter) and of calcium chlo- ride (0 to 86 g. per liter) were used in making up the Ringer’s solutions. Each experiment was performed at least in triplicate. The pH of the plasma-solution mixtures varied from 7.61 to 7.65 ; the pH of the Ringer’s solutions from 7.75 to 8.99. Roffo concluded that potassium favors the growth of normal and malignant tissues, whereas calcium hiders growth. Illustra- tive pictures are given. Koerbler (1925) inoculated fragments of a sarcoma into several series of rats and then injected the animals with mnoentrated solutions of sodium chloride and of “dipropanoldiphosphite de chaux (Gtaurol).” The doses were just under the lethal dose for rats, and the injections were performed regularly and repeatedly. Yet no differenoe in the growth of the tumors in the animals treated either with the sodium or the calcium salt waa obaerved as compared with oontrols. Koerbler stated that oliniaal obeerva- tions on the influence of caloium and sodium ions in weer do not appear to be aodrmed by laboratory experiments. Little is given in his brief article in the way of detailed data. Rohdenburg (1925) gave some data on the mineral content of blood, of tissues, and of various rat tumors, and attempted to cor- relate the number of takes with the relative salt content. The data were presented more as illustrative of his theory than aa con- vincing evidence. According to this theory cancer is due to a disturbance of the salt metabolism of the aell, as a result of which the cell becomes hypermineralized in comparison with the sur- rounding body juices. Rohdenburg suggested that attempts be made to increase the sodium content of the blood stream. As stated previously, Jones and Rourke (1927) could not oor- roborate Rohdenburg’s theory. In a crareful investigation they found that the sodium content of the blood plasma of mice bears no relation to susceptibility or resistance to transplantable mouse cancer, nor to the presence or absence of tumors. Cherner and 8hafiroff (1928) injected parathormone into rats bearing Flexner caroinomas. The serum caloium was inoreased 5 mg. per cent, and the calcium content of the tumors in the 9 treated animals was twice as great as in 4 controls. The tumors of the parathormone-treated rats grew more rapidly than those of the controls. Rats bear@ sarcoma 10 were treated similarly, but the calcium content of 6 such tumors wa~not inoreaeed over that of 5 oontrol tumors in untreated rats. The authors condude that by means of parathormone “The calcium content of F.R.C. has been increased but no inhibitory effect due to calcium was effected. ” Goldfeder (1928) found that calcium chloride exerted a retard- ing effect on tumor growth and suggested that this occurred pos- sibly because the calcium content of the tumor tissue was raised until it equalled that of normal tissue. In an extensive investiga- tion the effect of calcium chloride treatment on mouse carcinoma and on Rous chicken sarcoma was studied. Analyses for calcium were made of the tumors and of healthy tissue. The data have SODIUM, POTARSIULI, OALOIUM AXD MAGNEEITJM IN OANOER 968 been presented on pages 931 to 933 and in Table I. It was found that in mice calcium chloride increased the calcium content of the tumors and caused them to grow more slowly. In chickens also, administration of calcium chloride increased the calcium content of the tumors. While no striking effect on tumor growth was noted, the treated hem seemed to be better off than the untreated ones and tumor growth appeared to have been somewhat retarded. Tu- mors developed in all cases, but growth was delayed five or six days in animals receiving calcium chloride as compared with the controls. Goldfeder concluded that calcium has a retarding effect on tumor growth, but not a pronounced one. The inactivation of the Bous chicken tumor agent by means of caloium compounds was studied by Lewis and Andervont (1928). They tested the effect of various calcium compounds on extracts of this tumor. Addition of calcium oxide, calcium mono-basic phos- phate, or calcium hexose-phosphate in small amounts was found to inactivate the agent, which was also inactivated by plaster of Paris (2 per cent). Calcium carbonate was effective only when a very large amount was added. Calcium chloride and calcium lactate were not effective. The pH of the solutions used is not stated. Langfeldt (1929) reported that in mice on a diet deficient in po- tassium, transplantable mouse sarcoma 37 ceases to grow, and fre- quently disappears. After such regressions the mice are immune against a second transplantation with this sarcoma but not against transplantation with Ehrlich carcinoma. He found that a diet de- ficient in calcium has no influence on the growth of sarcoma 37, and retards but slightly the growth of the Ehrlich carcinoma and of tar carcinoma. Langfeldt's conclusions are based on the results of 364 transplants, but no details are available since his findings have been published only in a brief abstract. The effect of calcium chloride and of potassium chloride on tu- mor-bearing rabbits and on rabbits in which inflammation had been produced was investigated by Tokunaga and Nagaoka (1929) in a oontinuation of comparative studies on cancer and inflammation. Inflammation was produced with lycopodium and with kaolin. The tumor employed was Kato '8 transplantable rabbit sarcoma. A 3 per cent calcium chloride solution and a 1 per cent potasRium chloride solution were injected intravenously and also subcutane- ously. The calcium chloride injections produced no change in the size of the infiammatory nodule as compared with control rabbits. The potassium chloride treatment, however, caused a great in- crease in size. Intravenous injection of calcium chloride in the tumor-bearing rabbits produced larger tumors than in the controls. Similar injection of potassium chloride produced rapid growth of the tumors. 970 Y. J. 8- Tokunaga and Nagaoka concluded that, both in inflamed tissue and in tumors, calcium hinders growth and promotes necrosis, whereas potassium stimulates growth. However, only 12 tumor- bearing rabbits were used in all: 5 for calcium chloride, 5 for po- tassium chloride, and 2 controls. The effect of salts on the metabolism of tumor tissue was studied by Nakamura and Suzuki (+1930). They concluded that calcium and magnesium had no effect on glycolysis of tumor tissue, while potassium increased glycolysis. No details are given, de Baadt (1930)fed 10 young mice with an alkaline, potsssium- rich diet relatively low in nitrogen, consisting of whole milk, white bread, and potassium citrate. In about six to eight months half of the mice had tumors: 3 had lung carcinomas; one had tumors at the shoulder; and one had abdominal tumors. According to de Baadt’s theory, in mice that do not develop cancer the cells are in- sensitive to potassium. He recommended that an acidic, potas- sium-poor diet be used therapeutically. His ex*riment waa not controlled by having mice of the same strain kept for the same length of time on a stock diet. Zadik (1930) tried to reduce the potassium content and to raise the calcium content of animals by feeding a potassium-free diet for periods up to ten weeks. The diet was well taken but led to loss in weight in eight weeks and to death in about nine weeks. Tumor growth was stimulated rather than hindered. Also, large numbers of mice and rats were given sublethal doses of sodiumwbalti-nitrite to precipitate potassium; this compound was given by the subcu- taneous, intravenous, and other routes. Successful results, as far as inhibiting tumor growth was concerned, were not obtained. An attempt was also made to raise the calcium content and to lower the potassium content by injwting a special hormone preparation; the tumors were not influenced by this treatment. Zadik’s con- tribution is a condensed report of five years’ work. Beaause al- most all of it led to negative results, the author has omitted all de- tails, tables, bibliography, etc. It apparently represents the work of a careful investigator. Marqu6s (1930,1931)stated that cells deprived of ionic calcium undergo a change in properties which results in a greater aptitude for indefinite proliferation. No data are given. The comparative effects of calcium chloride and of parathyroid hormone on tumor growth were studied by Paik (1930,1931). He inoculated rats with Flexner-Jobling carcinoma and divided them into the groups desoribed below. “The body weights and tumor growth were measured in ten animals of each group before and for three weeks after transplantation, and each group was compared as to these details with its controls.” The control rats increased in weight at a regular rate. SODIUX, POTABSITJX, ommly AHD XAGNESIUY IN OANOEB 971 Group I received parathormone. The test rats increased in weight at about the same rate as the control animals. The tu- mors, however, grew more rapidly and were definitely larger than the control tumors. In Group II both parathyroids were extirpated. There was some loss in body weight instead of the gradual increase noted in the controls. For a while tumor growth was slightly slower, but later it became as rapid as in the controls. In Group III parathyroid glands removed from healthy rats were transplanted into the buttocks of test rats. The bpdy weight increased at about the same rate as in the controls. For about two weeks the tumor grew somewhat faster than in the controls ; after two weeks the growth rate was about the same in both groups. The rats of Group IV were given injections of a 2 per cent cal- oium chloride solution into the abdominal cavity, 0.25 C.C. per 50 grams of body weight being given daily beginning with the day after inoculation. The body weight decreased. “Tumor growth was markedly disturbed, diminished, or even absent. ” Paik concluded that inorganic calcium, as in the form of cal- cium chloride, has an inhibitory effect on tumor growth, while parathyroid hormone, on the other hand, stimulates the develop- ment of rat carcinoma. Aocording to Paik “Fujinami states that calcium or sodium aots on tumor cells not directly but indirectly, aad Kimura that the &ium ion has no direct influence on the growth of tumor cells, though it may have an indirect effect.” The original papers of these two authors could not be located. McDonald (1931) in a recent review states : “In our own lab- oratories, we have often found that, in spontaneous mouse tumors, calcium salts have an inhibiting influence on the growth of the tu- mor.” No details are given. In summarizing his point of view he states “In order to have a cure of cancer, conditions must be produced which will: . . . (4) produce a calcium-like effect, (5) reduce or prevent the potassium-like effect.” Crile, Telkes and Rowland (1931) claim to have obtained re- gression of tumors by the use of sodium chloride. They measured the electrical conductivity, capacity, and potential of tumors, and also performed “ionization” experiments. They say, “we are performing a series of experiments in which an attempt is being made so to ionize the cancerous tissue as to change its sign of charge. Various salts have been tried, but thus far Rodium chlo- ride has been most effective. This research is still in progress, but in 24 cases repeated and prolonged ionization has permanently changed the sign of charge of the cancer; that is, has given it the same sign of charge as the normal tissue. Coincidentally with the 972 M. J. 8HEA.B ahanging potential the tumor has diminished in size in 14 cases and in one instance has entirely disappeared. ” The experimental prwedures are not described ; no details are given. Jinguu (1931)inoculated Kyoto transplantable rabbit sarcoma into 6 batches of rabbits. One batch served as a control for the other groups, which received intravenous injections of salt solu- tions. The injections were begun eight days after transplantation, and were given daily; 1 per cent solutions of sodium chloride, magnesium chloride, calcium chloride, potassium chloride, and lithium chloride were employed. It was concluded that magnesinm “amelerated most intensively the growth of tumor,” that lithium chloride and potassium chloride accelerated it more or less, and that sodium chloride and calcium chloride inhibited growth. The number of animals used is not stated. Lshear (1933) gave a number of calcium salts to mice prior to and subsequent to inooulation with carcinoma 63 and sarcoma 180. Caloium chloride, calcium lactate, calainm carbonate and calcium citrate were given, alone and in addition to various organio sub- stances. The salts were incorporated with the diet in some ex- periments and dissolved in the drinking water in others: various concentrations of these salts were given. In other experiments he administered calcium lactate and calcium gluconate to mice prior to inoculation with sarcoma 180 and continued the treatment after inoculation. The salts were given intravenously, subcutane- ously, and intraperitoneally. Although at times it appeared that a slight retarding effect on tumor growth was obtained, consistent retardation was not noted on successive repetitions of these ex- periments. More than 1,200 mice were used in this investigation.

V. EFFEUTOF MAOWESXUM Because of the importance ascribed to magnesium in cancer by numerous authors, the r61e of this metal is considered separately. In previous seations occasional mention has been made of mag- nesium in connection with tumor or blood analyses or the adminia- tration of various salts. In addition to such more or less inci- dental work. on magnesium, there have been many attempts to demonstrate that magnesium is of benefit in cancer. Apart from the work of. Gaucher and Herscher (1899), who gave one case report, the articles dealing with magnesinm are of reoent date. In the case reported by Gauoher and Herscher con- siderable improvement was noted in a ease of epithelioma of the lip after treatment with cautery and application of magnwium chlorate. In a brief summary of his clinical experiences of eighteen yeare, BODIWH,POT~H, OAI,O~M AND ~GNIWM IN OANOEB 073 Begnault (1915-16,1918) reported that he had been using arsenic add in the treatment of superficial epitheliomas and that the best results were obtained when, in addition, magnesium silicate was given. The good results had first been attributed to the silicate, but in view of the recent work of Robin, of Dubard and of Delbet, Begnault now ascribed the beneficial action of the treatment to the magnesium. He therefore administered magnesium hydroxide and magnesium silicate twice daily to all patients with inoperable can- oer. For papillomas and superficial epitheliomas he oonsidered the results perfect ;in inoperable cancer the results were encourag- ing in that he obtained arrest and even regression, together with an improvement in the general condition. However, he gives no details. The number of patients treated is not stated, nor is men- tion made of control cases not receiving magnesium. The report of Dubard (1918), though brief, is somewhat more detailed. He gave 8 to 12 grams daily of magnesium carbonate to patients operated on for cancer, and obtained 24 per cent of rela- tively long survivals. He gave statistics for 1,689 cases of surgi- d intervention between 1901 and 1916. His theory is that, with advancing age, calcium is substituted for magnesium in the tissues, with the result that cancers develop. He therefore recommended that treatment with magnesium be used prophylmtically m well m ouratively. He also suggested the addition of phosphoric acid, sodium ffuoride, and manganese, as they appear to aid in the as- nimilrrtion of magnesium I Tbe following year Dubard (1919) again gave statistics of his results with the combination of surgery and magnesium salts and again stated that the assorted inorganic salts mentioned above ap- peared to improve the efficmy of the treatment and to aid in mag- nesium “remineralisation. ” He concluded that his seventeen years of clinical observation had shown the value of administration of magnesium. He found it inapplicable, however, to rapidly de- veloping mncer, and declared it unfair to condemn the method on the basis of results obtained in cases which were already terminat- ing fatally. In a still later paper Dubard (1920) repeated his hypothesis that in cancer the body is poor in magnesium. He tried oral, submtaneous, and intravenous injections of varions magnesium preparations, including metallic magnesium, colloidal magnesium oridea, organic magnesium salts, and inorganic magnesium salts. His best results were obtained by oral administration, for long periods, of a combination of magnesium salts and organic comb pounds containing phosphorus. Amording to Dnbard, magnesium exerts no local effect on the tumor, but its administration causes the “soil” of the body to re- turn to normal and the cancer consequently recedes. Dubard ad- mitted that he had no evidence which dehitely demonstrated this fact and stated that only a patient and methodical study of the mineral balance could establish it. However, he himself did not attempt any such patient and methodical study. He also concluded that phosphoric acid, especially in organic combination, appeared to be the fizafeur tlectif of the magnesium ion. No data are given in support of this theory. Dubard presented 12 case reports. He found that sarcomas were not benefited by his treatment ; only epitheliomas were influ- enced. The diagnoses were made on the basis of clinical and his- tological examination. About 40 per cent of the cases diagnosed as cancer were said to show improvement, generally and locally, or an impressive retardation of the process. Although Dubard recog- nized the weakness of the evidence he adduced on behalf of mag- nesium therapy, he nevertheless recommended it in the treatment of human cancer. Dubard’s (1931) most recent contribution contains no data of either experimental or clinical nature. He still considers mag- nesium, together with calcium and phosphate, to be of value in cancer, but he presents no data in support of this belief. 8timulated by the claims of .Regnault and of Dubard regarding the value of magnesium in cancer, Itami (1919) injected magnesium chloride intravenously into mice bearing Bashford’s carcinoma 63, Bashford’s carcinoma 206, and spontaneous carcinomas. No beneficial effect whatever was observed in 175 mice. ‘In fact, with carcinoma 63, the tumors were larger in the treated mice than in the controls. Itami concludes : these experiments cast serious doubt on the value of magnesium in the treatment of human neo- plasms. In fact, the aseertion may be ventured that this element will prove to be useless.” In agreement with Itami’s findings, Katase, Kawamura and Mizutani (1920) found that treatment with magnesium resulted in an increased number of takes of rat sarcoma (see page 969). Similarly Hoshino (1922) found that magnesium chloride emhurrccd the growth of transplanted rat sarcomas. He also reported that calcium interrupted the growth of this tumor and that a mixture of magnesium chloride and caldum chloride prevented its growth (see page 962). In the elaborate feeding investigation of 8ugiura and Benedict (1922) it was found that oral administration of magnesium car- bonate and of magnesium chloride had a slight but distinct CIC- celerutirrg effect upon the development of Flexner-Jobling rat car- cinoma. These investigators say: “This influence must be due to the magnesium ions, since neither carbonate nor chloride ions were found to possess such action.” SODIUM, POTASSIUM, OhLOIUM AND MAGNEBIUM IN OANOEB 975 Amording to Uei (1926) minimal doses of magnesium promote the growth of rat sarcoma in vitro and permit migration of the ma- lignant cells from the explant. With increasing doses, an inhibi- tion of growth is noted. No details are given by the author in his abstract. Beding (1923) studied the effect of magnesium on tar cancer in mice. Four groups of 5 mice each were employed. In Group I, the control group, 2 mice died because of the ill effects of the tar- ring. In the 3 surviving mice cancer developed between the 58th and 69th days. Group I1 received subcutaneously a piece of metallic magnesium. Four mice survived. The appearance of tumors was delayed for one month longer than in the controls. Group III received subcutaneous injections of “Mg colloidal.” In none of the mice in this group did tumors develop. Group IV reoeived subcutaneously a mixture of caloium and magnesium sulfate. Two mice died at the outset. None of the 3 surviving animala showed tumors. Reding realized that an inadequate number of experimental ani- mals had been employed. He stated that the results obtained in the treated mice were characteristically different from those ob- tained in the untreated mice, and that it was only the insufficient number of animals which prevented definite conclusions. He also stated that experiments on a larger scale were in progress. Iceding and Slosse (1923) determined the urinary and fecal ex- oretion of magnesium and calcium in man before and after ad- ministration of magnesium sulfate. Thev also implanted metallic magnesium under the skin of normal rabbits (the number of ani- mals used was not stated). It was found that administration of magnesium resulted in increased magnesium elimination in the urine and especially in the feces : the calcium excretion paralleled that of magnesium. One normal person was studied and one pa- tient with cancer ;one experimental period was investigated in each cam. Reding and Dustin (1923) reported the biopsy Andings on pa- tients who had been given intramuscular injections of magnesium sulfate by Slosse and Reding (1924). Thev reported that me- tastases of breast carcinomas rapidly diminhhed in volume after several injections and that there was also evidence of a diminution of the volume of neoplastic tissue. Involution of the metastases resulted in a general improvement of the condition of the patient. In this one-paqe summary. Reding and Dustin five no data. The number of individuals studied is not stated. In the absence of such information it is impossible to decide how much confidence is to be placed in their conclusion that their results have demonstrated, “avec toute Z’ivideme dksirable,” that carcinomatous metastases, and apparently certain sarcomas, are sensitive to magnesium. 976 Me J. BEEAB Apparently Beding abandoned the idea of repeating the mag- nesium experiments on a large number of snimals, for in a later communication Slosse and Reding (1925) say: “Si nous avons expirimed6 sur l’homme, et cette exp6rience nous a paru l6gitime et ntkessaire, c’est parce que l’6tude du cancer exp6rimental ae peut fournir de aonclusions rigoureusement applicable au cancer spontan6 de l’homme.” They injected salts of magnesium and of heavy metals into patients who had been given up by surgeons and radioIogists because the tumors had begun to metastasize freely. In 45 per cent of the cases a more or less transitory improvement was noted. In 7 to 10 per cent of the cases, decided improvement, lasting for many years, wae obtained. Rbum6s of 15 cases are given. In this paper Slosse and Beding described their technic so that physicians might be able to use it. Magnesium sulfate was given in three waya: (a) intramuscular injection of a 20 per cent solu- tion; (b) intravenous injection of a 10 per cent solution; (c) by “ionisation,” using compresses soaked in a dilute solution. Cop- per and lead preparations and a colloidal magnesium preparation were also given; all three metals-magnesium, copper, and lead- were administered to the same patients. Although they considered it too early to make more than a preliminary announcement, tbe authors considered their clinical investigations to be sutficiently thorough and confirmative of their animal experiments to warrant their conclusion regarding the action of ions on neoplasms. Another author who has written enthusiastically on the virtues of magnesium in cancer ie Delbet. In one paper, Delbet and Palios (19284 stated that after two passages in mice treated with mag- nesium chloride the cancer cell was modified and that the tumor loet, at least in part, its capacity for successful grafting. They bad injected 1 C.C. of a 12 per cent solution of magnesium chloride subcutaneously into mice. These injection# were given for four or five days before inoculation and were continued after inoculation with a rapidly growing mammary adenocarcinoma. In one series of animals a solution of magnesium chloride was incorporated in the food. It was found in the preliminary experiment that the ix&&ions had no effect on the number of takes, but had an obvious effect on the development of the grafts. Apparently 5 test mice and 4 con- trols were used in this experiment. In the second esperiment, 6 mice were employed. In one animal the graft failed to take and a second mouse escaped. This left 4 mice: 2 controls inooulated with untreated tumor, and 2 test mice inoculated with a tumor from a mouse that had been treated with magnesium chloride. The ad- ministration of magnesium chloride was continued in the 2 test SODIUM, POTASSIUM, OALOIUM AND XAQNBSIUM IN OANOEB 977 mioe. The control tumors grew much larger than the test tumors. In a third experiment, one of these tumors from a magnesium- treated mouse wa8 inoculated into 6 animals. No take8 were ob- tained in 4 mice; in the other 2 the tumors receded after twenty days. In contradistinction to others, Delbet and Palios used only the chloride. They insist that the choice of the salt is of very great importance, for the anion can play a significant r61e. Delbet and Palios were well aware that the number of animals employed was too small. This limitation was not one of choice, however, but was due to restricted facilities, they stated. 8hortly after this report appeared, Delbet, (todard and Palios (1928) announced further findings regarding the beneficial effect of magnesium. They implanted sterile stones in the gallbladders of a dozen guinea-pigs. Two of the animals died. Of the 10 survivors, 5 were kept as controls and 5 were treated with mag- nesium ohloride. After a period of about four months, the animals were autopsied. Gross examination of the gallbladders showed that the controls had large lesions while the magnesium-treated animals had only insignificant lesions. Microscopic sections showed that in 2 of the 5 control animals the lesions were cancer- ous; in a third animal the diagnosis was doubtful. In the mag- nesium-treated animals the sections showed no evidence of cancer. The malignancy was not tested by transplantation. In spite of the admitted weakness of the evidence the statement is made that theee experiments appeared to justify the conclusions previously formulated by Delbet, namely, that the regular use of magnesium halides can hinder the genesis of cancers and conse- quently can diminish the number of cancer patients. Another paper by Delbet and Palios (19283) reported bene- f3oial results with a mixture of magnesium halides in mice bearing sarooma and 3 different epitheliomas. The mice were fed bread dipped into a solution containing magnesium chloride, magnesium bromide, magnesium iodide, and magnesium fluoride. In addi- tion, suboutaneons injections of about 0.5 C.C. of this solution were given daily, The controls received bread dipped in water. This treatment is said to have retarded the growth of the tumors and diminished their virulence, for the authors failed to get successful bes after two or three passages through magnesium-treated mioe. These new experiments were held to confirm previous find- ings and previous conclusions. Consequently the recommenda- tion was made that patients be saturated with magnesium prior to operation. In the most extensive experiment 3 test animals and 3 controls were employed; in other experiments only 2 test animals and 2 controls were used. Furthermore, the results were not olear cut. 978 YaJ. SW In still another paper Delbet (1928) recommended the adminis- tration of magnesium salts as a prophylaxie for oanaer. The data and the photographs that appeared in the articles with Palios are supplemented here by the citation of a few cases of haman mcer. Delbet (1930) next reported his results in 2 rabbits with tar cancer, treated by daily injections of “delbiase,” though he fails to state what magnesium compounds enter into the composition of this preparation. He stated that the magnesium-treated animals were much better off than the 2 controls; cure was claimed in one case and regression in the other. Delbet and his oollaborators have interested themselves in a number of aspects of magnesium therapy. Delbet and Franicevic (1930,1931) studied urinary pH in 71 eubjects, approximately one- third of whom were healthy persons, one-third non-cancer patients, and the rest cancer patients. They concluded that in cancerous individuals there is a definite tendency for the urine to be alkaline, whereas in non-cancerous individuals the urine is usually acid. Treatment of the cancer patients with magnesium halides caueed the alkaline urine to become acid. The antagonistic effect of mag- nesium halides on cancer was ascribed, in part at least, to the acidosis produced. In commenting on this work, Cavers (1931a) says: “The authors conclude that their results confirm the view of &ding and Slosse that alkalosis is an important factor in the etiology of malignant disease. [It may be pointed out, however, that more than half (12) of their 22 patients showing alkaline urine had can- cer of the gastro-intestinal tract, in which there occur metabolic disturbances which would account for a change of the body fluids to the alkaline side and which, since they occur in various other dis- eases of this tract, have probably nothing to do with the causation of cancerous or even precancerous conditions, as is suggested by the writers mentioned.] ” In another contribution of Delbet and Palios (19314 additional evidence was presented in support of the thesis that magnesium is of value in cancer. They painted 12 rabbits with tar; 6 of these were given magnesium halides and 6 served as controls. Only 5 survived long enough for the development of cancer : 2 controls, both of which had cancer, and 3 magnesium-treated rabbits, none of which had cancer. The same data were presented in another article (1931b). According to Cavers (1932), “This is in one re- spect an improvement on a previous report by the senior author, when an’even smaller number of rabbits was used in hi0 ‘mag- nesium cure ’ experiments.” Stimulated by Delbet’s work, Zimmern and Wickham (1928) used magnesium sulphate in 5 patients bearing assorted inoper- 8ODmM, POTA88mM, OALOIUM AND MAQXEEIUM IN OANCIBB 979 able cancers, who had been abandoned by both surgeons and radiologists. A positive electrode was dipped in a conwntrated solution of magnesium sulphate and applied to the cancer or to a region near it; the indifferent electrode, wet with water, was so plaoed that the neoplasm was in the strongest portion of the elec- tric field. The current waa varied from two to ten milliamperes and given for from twenty to sixty minutes. In 3 of the 5 cases this treatment is said to have produced a temporary improvement after each administration. Kotzareff, de Morsier and Morin (1928) reported that tumors in white mice are favorably influenced by magnesium chloride. A transplantable epithelioma waa studied in 49 mice, of which 26 were controls and 23 received a 4 per cent solution of magnesium chloride orally. It was administered each morning as a drink. The 23 test mice were used in 6 successive transplantations. After the third passage, the tumors in the magnesium-treated animals were smaller than the controls, but were atill transplantable and were still rather large, even in the sixth passage. In some cases the tumors in the magnesium-treated mice grew even more rapidly than the controls at first. Since 23 test mice were used for 6 passages, apparently 4 test mice were used in each experiment. On the basis of such experi- ments the authors concluded that their work codirmed Delbet’s theory and that it showed that magnesium halides act advantage- ously in precancerous conditions. Marullaz (1929) gave magnesium chloride daily to 3 rabbits by the oral route. He then tried unsuccessfully to induce tar oancer in these animals. In discussing this paper, Delbet stated that he was aware of the scantiness of the experimental material and of the lack of controls, but that he believed the ‘work was significant. He added a report of some of his own work in which 4 rabbits were tarred; 2 were given magnesium chloride sub- cutaneously and 2 were controls. Of the controls, one had a papil- loma and the other a cancer; of the test rabbits, one died at the beginning of the experiment and the other one was rapidly and oompletely cured. Though recognizing the weakness of his data, Delbet considered that, taken together with the work of Marullaz, the evidence was so convincing “que je rephte comme un refrain: la saturation de l’organisme par le map6sium exerce une action onratrice sur certaines l6sions procanc6reuses et une mtion fr6na- trice sur la canc6risation. ” The following year Marullaz (1930) concluded that the appear- ance of tar tumors in rabbits is retarded and that the development of such tumors is hindered by treatment with magnesium chloride. 980 M. 3. BEMB He repeated his previous work, using 8 rabbits this time; he also gave magnesium chloride to 3 rabbits already bearing advanced tar tumors. He stated that magnesium does not cure existing tumors but it may cause a slight regression. In 2 albinos magnesium did not produce a favorable effect; in fact the tumors were even “pr6coces.” No mention is made of controls. Lemay (1930) discussed recent work on magnesium and claimed that it oonfirmed his diastatio theory of wncer, but ofered no data. The observation of Delbet and Franicevic that administration of magnesium halides to cancer patients causes alkaline urine to beoome acid waa disoussed by Oliviero (1931), who suggested a possible chemical mechanism for this reaction (see Caver’s com- ment, quoted on page 978). Another recent paper recommending the use of a magnesium preparation is that of P6ricaud (1931). He recommended the use of “Hopogan, ” apparently a proprietary preparation of mag- nesium peroxide, not only in the treatment of cancer, but also as a preventive. No caae reports were presented ; no experiments with magnesium were reported. In a review of work done in his laboratory, Maisin (1931) stated that one of his students, Estaa, found that magnesium sulfate given in suitable doses is capable of delaying the appear- ance and the evolution of tar cancer in mice. Honor6, another student, obtained analogous results with magnesium chloride. No details or data are given in this review. Dufour (1931) cited 3 cases in which magnesium chloride and pyoformine (apparently a proprietary antiseptic preparation) were given. One patient died nine months after the treatment was begun; another was still alive but in bad shape; in the third the treatment was sumessful. Of the second patient Dufour said: “Cependant le progrhs de la maladie se font avec une lenteur qui a dbu tous lea pronostios.” To quote Cavers (1931b), Dufour “admits that spontaneous amelioration, partial regression and slowing down of growth frequently prolong the course of inoper- able malignant disease, but is inclined to attribute considerable importance to the ‘magnesium cure’ described.” Another investigator, who reported that magnesium did not check tumor growth, is Bolaffi (19300). To mice bearing an adeno- carcinoma she gave magnesium chloride both orally and subcu- taneously. Except in a few instances, the treatment with mag- nesium did not lead to a prolongation of life in the tumor-bearing animals over that of untreated controls. It hindered growth in only a few cases and, as a matter of fact, at times appeared to favor tumor growth. In one group the tumors in the treated mice were obviously Zarger than those of the control mice. The tumors BODIWY, POTASSIUY, C~ALOIUMAND YAWIBSIUX rx OABOBIB 981 were analyzed for potassium, calcium, and magnesium. Treat- ment with magnesium chloride did not produce any change in the magneeium content of the tumors. In a few instanoee the po- tassium: calcium ratio decreased, but these caeee were not the ones in which an inhibition of growth oclourred. In view of the fact that the results obtained were at such un- expected variance with those reported by Delbet, hietologio ex- aminations were made; no difference between the treated and untreated tumors was found. Bola did not find any evidence which would indicate that magneeium has a favorable effect upon the reduction of growth energy of malignant tissues. Apparently only 31 mice, including controls, were employed; each group con- tained about 4 mice. Since she had found that magnesium chloride did not have any profound action on tumors, Bold (1930b) in a subeequent study gave orally to mice a mixture of magnesium phosphate and tartaric acid in one experiment and a mixture of magnesium iodide, mag- nesium .phosphate and tartaric acid in another experiment ; the tumor employed was Ehrlich ’s adenocarcinoma. She concluded that magnesium does not have a significantly beneflcial effect on the development of this mouse carcinoma. Apparently only 17 animals, including controls, were employed. In both of these etudies the comment of Cavers (1931~)is pertinent : “The number of animals used . . , and of tumours analyzed seems too small for any conclusions to be convincing.” At the Buffalo meeting (September 1931) of the American Chemical Society, McDonald and Itami (1931) reported that mag- nesium is without effect upon spontaneous adenocaroinoma of the breast in mice. They injected solutions of magnesium gluconate intravenously into a strain of mice which had a high percentage of spontaneous tumors ; no beneficial results were obtained. The effect of magnesium on the growth of a large ephdle-cell sarooma in albino rate was studied by Cilotti (1931). The control rats received water containing less than 0.004 per aent magnesia, while the test rate were given water containing 12 per cent mag- nesium chloride. Several series of experiment8 were wrried out, in which the time of administration of magnesium chloride prior to tumor inoculation was varied. The ingestion of this large amount of magnesium chloride had no influenae on the percentage of takes or on the rate of growth of the tumor. Still another investigator who reported that magnesium does not retard tumor growth is Jinguu (1931). He found that in- travenous injection of 1 per cent magnesium chloride solution into rabbits bearing Kyoto transplantable ~arcomaproduced an accelerartiols of tumor growth as compared with the control ani- male. (For additional details of Jinguu’s experiments see page 972.1 Thomas and Ereitmann (1931) reported that a preparation of magnesium halides was efficacious in the treatment of thyroid epithelioma of salmonoid fish. This preparation was given to 15 fish, of which 11 survived; the growths disappeared. In 31 of 33 other fish not given the magnesium preparation, tumors unex- pectedly did not grow. This was ascribed to unintentional wash- ing of magnesium salts from the compartment containing the treated fish into that which contained these 33 fish. In other com- partments, tumors developed as usual in control fish. Lorenzetti (1931) stated that during pregnancy the mother’s store of magnesium is reduced and that consequently the mother is exposed to the danger of cancer. He therefore suggested that pregnant women be dosed with magnesium as a prophylactic meas- ure. According to Cavers, “The sole comment required is that some of the author’s premises have not been dehitely estab- lished. ” The importance attributed to magnesium by certain authors led others to investigate the correlation between the distribution of magnesium in soil, salt, food, etc., and the incidenoe of cancer. Aomrding to Marchi (1930), each province of Italy is served by a separate salt works ; he compiled government data on the amount of salt consumed and on the magnesium content of the salt in each district salt works. In one brief table showing the gross cancer death rate in each district and the magnesium content of the salt used, an inverse relationship was pointed out. Marchi stated that, while this alone would not be conclusive, taken in conjunction with the animal experiments of Delbet and of Marullaz it appears to have significance and shows that cancer is more prevalent where the magnesium content of the diet is low. Bernadini (1930) had previously studied the r6le of mapes- ium in green plants. Stimulated by the publications of Delbet, he attempted to correlate his own work with the medical and nutritional aspects of magnesium, especially with the function of magnesium in the human nervous system and in cancer. He accepted at face value the conclusions of those who claimed that magnesium is of benefit in the treatment and prevention of cancer, and disoussed the distribution of magnesium in plant foods. This paper contains no contribution to the chemistry of canoer. A striking correlation between the magnesium content of the soil and cancer mortality was reported hy Robinet (1930a, 1930b). He had previously published data for most of the departments of France which, he asserted, demonstrated that the greatest mor- tality from cancer was found where the soil was poor in mag- SODIUM, POTMIUM, OALOIUM AND MAGNEBIUM IN OANOEB 983 neaium.’ hsett (1929) stated that Robinet’s theory did not apply to Alsace-Lorraine and presented data in refutation of Robi.net ’8 thesis. Robinet thereupon extended his investigations to &ace-Lorraine, which had not been included in hie previous study. He gave maps which showed the magnesium content of the earth and corresponding maps for the cancer mortality of these departments. The maps showed a striking correlation. How- ever, not a single chemical analysis of the soil was reported. The maps were taken from geologic surveys and showed the distribu- tion of different types of geological formations, Robinet’s assump- tion being that such geological surveys give the magnesium content of the surface soil. His thesis would be far more convincing if actual chemid analyses of surface soil showed a similar correla- tion with cancer mortality. The same comment applies to his study of cancer mortality and geological formations in England and Wales, in which he (1931) reached the same conclusions as he did for Frame. A similar thesis was advanced for Egypt by Schrumpf-Pierron (19310,1931b). As the result of a statistical study, he concluded that the incidence of all kinds of cancer, and especially cancer of the gastro-intestinal tract, is very low in Egypt. He stated that malignant tumors occur with one-tenth the frequency that they do in Europe and sscribed this low incidence to the high magnesium content of the Egyptian soil. He concluded that the frequency of cancer is inversely proportional to the amount of magnesium in the soil. Dukes (1931), in commenting on Schmmpf-Pierron’s work, states: “No allowance is made for the difference in age constitu- tion of the population in Egypt and European cotdries, and the author’s statistics are open to criticism from many points of view.” A detailed analysis of Schrumpf-Pierron’s work was made by Brnmpt (1931a). In a critical study Brumpt pointed out that the alleged rarity of cancer in Egypt is a mere consequence of the inferiority of the statistical system of the country. He himself presented statistical data which, he concluded, shows that cancer is not infrequent in Egypt. This article also contains a defence by Delbet of Schrumpf-Pierron ’8 position. In another article Brnmpt (1931b) discussed the theses of Schrumpf-Pierron as regards the incidence of cancer in Egypt and the alleged value of magnesium salts, and concluded: (5l me semble inutile d’insister sur le r6le hypothetiqne que ces eels jouient en Egypte.” Finally, *Bobin& (10300) refem to thin earlier publication an: 6‘Terrsina msgn6depu et 0.nOer.” Etude aomperative pour la Frsnae. La Prophytds du Concur, Editew; Monfin ti Moret-s/-Loing. 984 IYI. J. SHHAB Bmpt (1931~)showed that oral administration of magnesium salts is without effect on the development of spontaneous canwr in white mice. Tumors developed in 27 of 89 magnesium-treated mice and in 24 of 134 control mice. The magnesium was given in amounts that were from 20 to 60 times the dose reoommended by Delbet as the preventive dose for man. VI. PHYBIOO-OHBMIOLU(STUD= In addition to quantitative analyses of tissues and of body fluids and to studies of the effect on tumor growth of the adminis- tration of salts of the common cations, investigations have been oarried out in which physioo-chemical technios have been employed in an attempt to throw light .on the various phenomena in whioh sodium, potassium, calcium, and magnesium are involved, Thus Clowes (1914) studied oil-water emulsions with regard to the effect of ions, including calcium and magnesium ions, upon such systems. This was followed by a study of the action of antagonistic elec- trolytes on emulsions and living cells, in whioh Clowee (1916) in- vestigated the effect of caloium, sodium and magnesium salts. Electrical Properties: In a brief report of further studies in which he determined the electrical conductivity of primary breast carcinomas of mice, Clowes (1917-18a) stated that these prelimi- nary experiments indicated that “cancer tissues are more permeable to ions than are normal tissues and that the permeability bears a definite relation to the speed of growth, rapidly growing tumors exhibiting a lower resistance than slowly growing tumors.” Simi- lar determinations with plant tissues and plant galls indicated that “plant tumor tissues are uniformly more permeable than normal plant tissues, the tumor tissues frequently exhibiting, under rea- sonably comparable conditions, a oonductivity more than twioe that of normal tissues.” No data are given. In another brief report Clowes (1917-186) summarized experi- ments in whioh arMcial emulsion membrwes were exposed to solutions of sodium chloride and of calcium chloride and the eleo- trical conductivity then measured. He concluded that ‘ The above data correlates admirably with the well-known fact that alkalis, salts of sodium, potassium, eta, promote the permeability of tissues while salts of caloium, magnesium, and other di- and triva- lent oations exert the reverse effect. Ale0 with the observations of Beebe and the writer regarding the high IC content and low Ca oontent of rapidly growing tumors and the low content of K and high Ca content of slow-growing or stationary tumors.” No data are given in this paper. Clowes (1919) subsequently reported briefly on similar studies of eleotrioal reaistanoe of Wcial emul- sion membranes and of tumor tissue. He found that sodium ohlo- EIOD~M,POTASSIUM, OALUIUM AND MAGNESIUM IN OANCYEB 985 ride inoreased and that calcium chloride’ decreased conductivity, These articles were preliminary notes. Waterman (1922) carried out an extensive series of electrical measurements on the resistance and polarization of normal and tumor tissue. The resistance was measured in the usual way, employing a Wheatstone bridge, using non-polarizable electrodes to hold the pieoes of tissue. The polarization was determined by inserting a variable self-inductance in the system to compensate for the phase difference produced by the polarization. The re- sistance (R)was given in ohms and the polarization (P)in milli- henries. For normal tissue treated with Ringer’s solution Waterman found that P/R -0.020; for malignant tissue similarly treated he obtained a value of 0.004. On substitution of isotonic sodium ahloride or potassium chloride for Ringer’s solution, no pro- nounced effect was obtained with normal tissue; however, when isotonio calcium chloride was used, a lower P/R ratio was obtained. Tumor tissue gave an opposite reaction, i.e., when treated with isotonio calcium chloride, it showed an increase in the ratio. In a subsequent paper Waterman (1923) reported further findings. In his earlier studies Waterman worked with normal and can- cerous tissues. In a later investigation (1938) he worked with normal and malignant cells, using a suspension of lymphocytes as typioal of normal cells. Cancer cells gave a value for P/R only one-fourth that given by normal cells; this paralleled the results obtained with tissues. These electrical properties were found to be affected by various agencies. The immediate effect of irradiation of cancer cells with x-rays was an increase in P/R. With lymphocytes this first effect was reversed : irradiation caused a reduction in P/R. The effect of calcium ions was similar to that of x-rays. In cqncer oells cal- oium produced an increase in P/R,whereas with lymphocytes it produced a decrease. Treatment of cancer cells with either cal- cium ions or x-rays caused them subsequently to behave more like normal cells, Thus after treating a cancer cell suspension with aalcium, irradiation with x-rays produced no further increase in P/R. Similarly, after irradiation with x-rays canmr oells be- haved more like normal cells on treatment with aaluum, Attention may be drawn here to the report of Rohdenburg and Krehbiel (1924) that seventy-two hours after irradiation ‘of a tumor with x-rays there was a decrease in the potassium content, whereas the potassium content of a control, non-irradiated tumor in the opposite axilla of the ~~amerat showed no suoh decrease. They studied three types of rat sarcomas and did three to six analyses on each. 986 M. J. BHEAB Crile, Telkes and Ro’wland (1931) stated that they had meas- ured the electrical conductivity, capacity, and potential of tamOrS. They also stated that “ionization” with sodium chloride resulted in a change in the sign of charge of the cancer (of. page 971). No data are given. 8urface Tedsiora: Bauer (1923) studied the surface tension of blood serum and discussed the bearing of calcium and magnesium on this physical property. Svehla (1926) stated that it is well known that blood serum in cancer has a surface tension lower than that of normal serum, and that extracts of malignant tumors have a lower surface tension than extracts of homologous healthy organs. Since, Svehla stated, addition of calcium produaes an increase in surface tension, he attempted to correlate the surface tension and calcium content of the blood serum of cancer patients. In the latter he found the surface tension to vary from about 48 to 55 dynes per cm.; in normal individuals it varied from about 63 to 65 dynes per om. In cancer patients he reported that the serum calcium varied from 14.2 to 22.0 mg. per 100 C.C. and in normal persons from 22.6 to 24.6 mg. per 100 C.C. He concluded that the lowering of the surface tension runs parallel with the decrease in serum aalcium. Inasmuch as Svehla used the well known Kramer-Tisdall method in determininq serum calcium, it is difiionlt to understand why he obtained such high values for calcium. The values ob- tained for surface tension may be correct but the correlation iR invalid beoause the calcium values are obviously wrong. Absorptiora amd Imbibition: According to Wolf (I 923), normal tissues have a fairly constant capacity for ahsorntion. whereaR cancer tissue has a strikinElv exaggerated capacity. He stated that cancer tissue, on immersion in salt solutions, can absorb three times the amount of potassium and twenty times the amount of calcium originally present. No details are given. There is a considerable amount of experimental evidence which supports the view that growing tissues in general and neoplastic tissues in partimlar are richer in water than non-proliferating tissues. In addition a number of investigations have been re- ported in which the avidity of malignant tissue for water has been studied by in vitro experiments. Thus, Lamitzki (1928) postu- lated that the rate of cell growth is dependent on the degree of imbibition of water by the cell and that this in turn is favored by potassium ions and lowered by calcium ions. He immersed pieces. of Flexner-Jobling rat carcinoma and of Jensen rat garcoma in various solutions and determined the metabolism according to Warburg’s method. For controls he used Ringer’s solutions free SODIUM, POTABIWM, OXIIWM AND XAONESIUM IN OANOEB 987 from potassium and calcium. He then observed the effect of add- ing potassium chloride. With carcinoma he did 55 potassium chloride experiments and 42 control experiments: with sarcoma, 14 potassium chloride and 14 control experiments. Rimilar results were obtained with both types of tumors: the addition of potassium chloride increased lactic acid formation. Addition of sodium chloride or lithium ohloride instead of potassium chloride did not produce this effect ; it waa not, therefore, due to increase in the ionic strength or in the osmotic pressure of the solution. Neither was it due to the chlo- ride ion or to a non-spWc effect of monovalent cations. It was due specifically, Lasnitzki states, to potassium. Addition of cal- cium chloride or magnesium chloride did not give similar results. Lasnitzki concluded that potassium has a specific effect on imbi- bition of water by tissues, and that this leads to increased meta- bolism which, in turn, produces increased growth. He represented this series of reactions diagrammatically as follows :

Potassium 3 Imbibition Capacity + Metabolism Growth. Rohdenburg and Bernhard (1929), in a report of experiments on imbibition ba oitro, state that malignant tissues suspended in certain salt solutions take on less weight than benign tissue. If suspended in a salt solution corresponding to the blood of animals bearing retrogressing tumors, malignant cells lose weight. Analy- sis of such a salt solution, after immersion of tumor tissue, shows no striking changes except that the tumor tissue takes up some sodium from the solution and gives up potassium to the solution. The solutions used were made up on the basis of earlier analyses of Rohdenburg et a2 of the blood of tumor-bearing and control ani- mals and of patients. As has been pointed out in an earlier sec- tion, however, the data gathered by these investigators are of little value, since the analyses were done on whole blood. In order to interpret experiments of this sort in the light of blood chemistry, the tissues should be immersed in solutions having the composition of plasma or of serum. In reporting the losses and gains in weight of the immersed tissues, and the changes in the composition of the sohtions follow- ing such immersion, Rohdenburg and Rernhard give no details, either as to the actual weight ohanges or the individual percentage change for each experiment. An extensive series of experiments by Magath appears to be a careful piece of work, with reneated checke and controls and with numerous variations. Mapath (1927a) studied the B vitro ab- sorption of water bv normal and cancerous tissues under the indu- ence of acid, employing two rapidly growing strains of transplant- 988 M. J. EHBds able carcinoma-strains “N,” and “N,” of mammary adenmr- cinoma of white mice. No essential differences were found in the imbibition capaoities of oancerous as contrasted With normal tismes except when the tissues were first treated with isotonic calcium chloride solution. Such treatment greatly diminished the imbi- bition of cancerous tissue, but incressed that of normal tissue. Only fresh, living tumor tissue, however, reacted with decreased imbibition; necrotio and older tumor tissue did not behave in the same way. The spleen and, to an even greater degree, embryonic tissue behaved like cancer tissue. These experiments showed extensive analogies to the electro- chemical findings previously reported by Waterman on the polar- ization-resistance ratio of tissues and on the effeot of calcium treat- ment on these properties. In a later study, Magath (19276) used a strain of rat sarcoma and rabbit liver in addition to the tissues previously used. He again found that cancer tissue imbibes less water when immersed in acid solution if it is first treated with calcium chloride, whereas normal tissue imbibes more water after such treatment. Further- more irradiation of the tumor in situ and of rabbit liver in the liv- ing animal produced a striking effect upon subsequent imbibition experiments. After x-ray treatment, tumor tissue imbibed less than non-irradiated tumor tissue when it was plaoed directly in acid solution. Thus the effect of x-rays was analogous to that of caloium chloride. finally, when irradiated tumor was treated with caldum chloride it showed a striking increase in imbibition on im- mersion in acid solution. Thus irradiation reversed the effect pro- duced by treatment with calcium chloride, and caused tumor tissue to behave like normal tissue. In further studies Magath and Kolomijetz (1930) repeated with Ehrlich’s mouse carcinoma, Jensen rat sarcoma, and chicken em- bryo tissues the experiments that Magath had previously per- formed with the mouse adenocarcinoma “N” of Dr. Petrow (Leningrad). These tissues gave results analogous to those ob- tained with tumor “N”; i.e., in general, treatment with calcium chloride reduced the amount of imbibition. This reduction waB reversed by irradiating the tumor in vivo with x-rays. Somewhat similar results were obtained with embryonic tissue. Further- more, irradiation of tumors in vitro gave results analogous to those obtained by x-raying i~vivo. Magath and Kolomijetz tried to correlate these effects with the calcium content of x-rayed and untreated tumors. They found that the calcium content of normal livers and of tumors varied widely from one individual to another. No dehite increase in calcium oontent was observed following irradiation. Even after immersion in calcium chloride solution the x-rayed tumors showed SODRJM, WTA~~IUM,UALUIUM AND IAGNEBIUM IN UANOEB 989 no distinct difference in calcium content from the control tumors. In these experiments the authors used carefully prepared tumor tissue, free from necrotic areas, and normal liver. The method of Pincussen-Cronheim was employed ; the tissues were ashed with nitric acid and perhydrol. This method is said to be accurate to 3 per cent. Tissue Respiratiow No attempt will be made here to review the extensive literature on tissue respiration. However Warburg’s (1931) statement regarding the influence of salts may be cited. He suggested that the action of salts on developing tissues may be due to “their influence on the oxidative reactions.” It is well known that marine animals soon die if placed in a pure sodium chloride solution of the concentration of sea water, but that the harmful effect of the sodium chloride solution is counteracted by the addi- tion of calcium ions. Warburg’s own experiments led him to the oonclusion that sodium chloride increases cell oxidation many fold : “In a pure NaCl solution the oxidations of the fertilized ovum in- crease so greatly that this increase alone quite sufficiently explains the poisonous aution.” Lasnitzki and Rosenthal (1929) studied the effect of omission of potassium and calcium from Ringer’s solution in determinations of lactic acid fermentation of tumor tissue. The respiration of, normal tissues, of Jensen rat sarcoma, and of Ehrlich mouse car- oinoma 5 was studied by Kisch (1931a, 1931b) with regard to the effwt of inorganic salts. This work was an extension of that of Liebowitz (1930), whose paper contains a discussion of previous investigations on the effect of salts on tissue respiration. Potassium and Physiological Radio-activity: Due chiefly to the work of Zwaardemaker, physiological importance has been ascribed to the radio-active properties of potassium. lgince potassium is but feebly radio-active, it had been aasumed that it exerted no physiological effect by virtue of this property. However, Zwaardemaker (1917-18), subjecting the question to experimental study, reported interesting findings. He perfused frog’s heart with various solutions and obtained the usual results: in the ab- mnce of potassium, the heart beat stopped; when potassium was added, beating was resumed. It was found that other radio-active rcabstances could be successfully substituted for potassium. Too little of the substituted substance had no effect, and too much was harmful; but when an amount was added which was equivalent in radio-active power to the radio-activity of the potassium, the same physiological effect was obtained as with potassium, i.e., the heart beat was resumed. Calcium was found to have an antagonistic effect: the greater the amount of calcium in the perfusion liquid, the greater was the concentration of radio-active substance needed to maintain the automaticity of the heart action. 990 M. J. BHEAB Since rubidium, uranium, thorium, radium, eta., were found to have the same physiological effect as potassium, Zwaardemaker concluded that the automatie regulation of heart action by potas- sium is due to the effect of its beta radiation. Zwaardemaker (1919-20) a160 found that heart beat was re- sumed, during perfusion with a potassium-free fluid, under the in- fluence of mesothorium radiation acting at a distance. In a later paper he (1920) summarized the results of years of experimental work on this subject. Analogous results were obtained (1922) with other muscles that exhibit automatic pulsation, i.e. frog’s esophagus, mammalian intestines and uterus. Zwaardemaker’s work is occasionally cited in discussions of malignant growth. It has also been mentioned in connection with the controversial subject of mitogenetic radiation. Lubbrecht ’s (1920) experiments, however, appeared to cast doubt on Zwaarde- maker ’s theory. Zondek ( 1921 ) ,also, could not drmZwaarde- maker’s work, and concluded that the physiological effect of po- tassium is not related to its radio-activity. Clark (1921) found “that rubidium aets as B true substitute for potassium, and that it can replace potassium in all tissues and that excess of rubidium acts Eke excess of potassinm; on the other hand, uranium and thorium can compensate to a certain extent for lack of potassium in the frog heart under a certain narrow range of conditions, but under all other conditions in the frog’s heart, and under all con- ditions in other tissues there is no evidence at all that uranium and thorium act as substitutes for potassium. ” According to Stoklosa (1925), the human body contains 30 grams of potassium, which emit 13,800 beta particles every second. In his review, Stoklosa summarized his hdings on the relation of radio-activity to respiration, oxidation-reduction, enzymes, carbo- hydrate metabolism, potassium, etc. Most of his work was done on plant tissues. Roffo (1924-25c) found that potassium favors the growth of normal and neoplastic tissue and suggested that the radio-activity of potassium may be part of the stimulating mechanism. Boffo and Landaburu (1925) reported that injection with rubidium chlo- ride solution resulted in an increase in the radio-activity of organs. Tumor tissue always showed a higher radio-activity than normal tissue, probably because tumor tissue absorbs more rubidium, they stated. Enayme Actiota: In connection with the enormous amount of work that has been done on the metabolism of malignant and of normal tissues, it may be pointed out that magnesium appears to be involved in some of the enzymatic processes concerned. Jenner and Kay (1931) recently reported that animal phosphatases are activated by magnesium, thus confirming the data of earlier work- SODIUM, POTASBIUM, OALOIUM AND MAGNESIUM IN OANOEB 991 era. Keeser (1931) studied the effect of calcium and of magnesium on various enzyme reactions. Activating effects were noted in some instances and inhibiting effects in others. Marques (1931) stated that many enzyme reactions are de- pendent on the presence of ionic calcium, and that calcium activates oxidases and proteolytic enzymes. No data are given. In a review of oarbohydrate metabolism and of growth, von Euler (1931) discussed work from his institute which showed that magnesium is indispensable aa an activator of certain empere- actions. As yet no definite relationship between the magnesium content of tumor tissue and its metabolism has been demonstrated. VII. DIEOUSSION From a perusal of the preceding sections it is obvious that muoh confusion exists as regards the r61e in cancer of the commonly m- ourring constituents, sodium, potassium, calcium, and magnesium. In evaluating the conflicting claims, the experimental evidenoe for and against the importance of each of these elements in cancer will be oonsidered separately. Furthermore, in order to avoid muddy- ing the waters, little or no attention will be paid in this section to work that is of doubtful significance ; speculations based upon scanty experimentation, data obtained by unreliable methods, and inadequately controlled clinical or experimental observations will not be oonsidered at length. Bodium Bodiulrr Code& of Tmor Tissue: In Beebe’s (19W)pioneer work sodium was obtained by difference ;it was not determined by direct analysis but was obtained by calculation. Clowes and Frisbie (1905) reported values for the sodium content of Jensen mouse carcinomas but did not mention the method by which they obtained their data. Waterman (1921) tried to determine sodium dirwtly, using the then recently devised pyroantimonate method of Kramer and Tisdall (1921). However, he could not get a pure reagent, and poor results were obtained so that “L’inconv6nient olaesique, que toutes les erreurs d’analyse se traduisent dans des erreurs sur la teneur en Na, se retrouve done dans ce travail, oomme chez lea auteurs prihMents, en sustrayant la quantit6 de . . . KC1 de la teneur totale en chlorures.” Thus, up .to and in- oluding the time of Waterman (1921, 1922) there were no reliable analyses of tumor tissue for sodium. Direct analyses were reported by Rohdenburg and Rrehbiel (1922,1924). The aata, however, are given in terms of wet tissue, and cannot be recalculated on the dry basis, since values for water 46 992 M. J. BmAR oontent are not given. No subsequent analyses for sodium have been published. In short, no information was found in the literature on the con- tent of sodium per gram of dry tumor tissue. Sodium Corderat of Blood: Before 1920 no accurate analyses of the sodium content of blood in cancer were made, chiefly because reliable micro-methods of analysis were not available. The first contribution of importance was that of Waterman (1921), who states: ‘“he Btude des rapports de Ca, K et Na dans le s6rum sanguin eat donc tout indiqu6e (chea des personnes normales et chez des malades attients de tumeurs), d’autant plus que dans ces dernier tempe seulement une microanalyse de ces constitnants dans le s6rum sanguin eat devenue possible et qu’i mon avis on ne peut, d’un point de vue m&thodologique,attribuer aucune valeur B des oommunications prMdents se rapportant B ce sujet.” Waterman reported that the sodium’content of blood serum is unchanged in cancer patient a. Theis and Benedict (1924), in a careful study, also found that the blood serum in cancer patients contains the normal amount of sodium. The data of Rohdenburg and Krehbiel (1922,1925)need not be considered beoause their analyses were performed on whole blood instead of on serum or plasma. The same comment applies to the extensive data of Blumgarten and Rohdenburg (1927). The assertions of Rohdenburg and his collaborators regarding the changes in the sodium content of blood in cancer led Jones and Rourke (1927) to investigate this point further. In an extensive investigation they found no signifwant difference in the sodium content of blood plasma of mice susceptible to an adenocarcinoma and of mice that were not susceptible. Furthermore, no changes from the normal sodium content were noted in the blood plasma of tumor-bearing mice, of non-tumor-bearing mice, or of non-sns- ceptible mice six days after inoculation with the tumor. The au- thors concluded that, “Since the sodium content is so relatively constant in such diverse groups as the normal mice, the tumor bearing, and those with active immunity reactions, we are led to believe that the sodium concentration alone cannot be the deter- mining factor in immunity or susceptibility. . . .” Regarding the work of Rohdenburg and Krehbiel, Jones and Rourke say: “The results of their analyses on blood, however, are difficult to interpret in terms of concentration of the different cations because the analyses weredone on whole blood without ref- erence to plasma volume percentages.” Jones and Rourke found that “there is much less sodium in the red blood corpuecles of mouse blood than in the plasma. . . . It is, therefore, quite pos- sible that the increased sodium content of the whole blood found by SODIUM, POTASSIUM, ChLaIUM AND MAGNESIUM IN CANOEB 993 Bohdenburg and Krehbiel may have been due to increased plasma volume percentage and not to increased sodium concentration of the plasma or cells.” As recently as 1929 Schepetinsky and Kafitin wrote: “Wir haben im Schrifttum keine Angaben finden konnen iiber den ge- samten Mineralbestand des Blutes bei malignen und benignen Tumoren. Die Befunde vieler Autoren iiber den Gehalt an Ca, K und P, auf GCrund verschiedener Verfahren an verschiedenartigem Material gewonnen, sind so divergent, dass es schwer ist, sich ein Urteil zu bilden iiber die vorgefundenen Veranderungen im Min- eralbestande des Blutserums bei Krebekranken. ” In an exten- sive investigation these investigators obtained data on the con- centration of sodium in the blood serum of a large number of fe- male patients and of normal women. They reported 44normal’7 values for serum sodium in patients bearing tumors, both benign and malignant tumors. In the majority of the cases the sodium oontent was between 200 and 300 mg. per cent. This was the value obtained in 73 per cent of the healthy women, although the normal value is 335 mg. per cent according to Peters and Van 8lyke (1931). Louros and Gaessler (1929) reported finding a marked lower- ing of serum sodium in 15 cases of carcinoma of the uterus. Uaes- sler (1930) obtained low serum sodium in 9 cases, both before and after irradiation. The sodium content of blood serum and of blister fluid was dc- termined by Pitts and Johnson (1930) in ti series of cancerous and non-cancerous individuals. In this careful study the sodium con- tent of these fluids was found to be the same in cancer patients and in normal persons. Treatmefit with Sodiwn: Cramer (1918)immersed an emulsion of mouse carcinoma 63 in a solution of sodium ohloride and then inooulated the cells into mice. He found that treatment with sodium chloride produced no detectable effect, whereas Rimilar treatment with calcium chloride produced a distinct inhibition in growth, Furthermore, the inhibitory effect of the calcium could be counteracted by subsequent treatment with the sodium salt. Analogous results were obtained by Sugiura, Noyes, and Falk (1921), who immersed Flexner-Jobling rat carcinoma in sodium ohloride solutions, Immersion for twenty-four hours had no effect, whereas immersion in calcium chloride solution resulted in complete inhibition of growth. When sodium and potassium chlo- rides were also present in the solution, the inhibitory effect of the calcium was not observed. The effect of oral administration of sodium chloride on this same rat carcinoma was studied by Sugiura and Benedict (1922). Neither the number of takes nor the development of the grafted tumors was affected by ingestion of sodium chloride. Similar negative results for a mouse carcinoma were obtained by Hiindel (1924), who also gave sodium chloride by mouth. Various concentrations of sodium chloride were employed by auraand Wada (19241,who immersed pieoes of Jensen rat sar- ooma, Flexner carcinoma, Rato rabbit sarcoma, and Kyoto fowl 8arooma in salt solutions and then inoculated the tumor tissue into animals. They found that the rate of growth of the tumors was not retarded by treatment with sodium chloride. In a paper entitled ‘‘A Theory of the Origin of Tumors” Roh- denburg (1925) advanced the conception that when the oell be- comes hypermineralized it grows and divides without cessation and thereby becomes malignant. He says: “the percentage of sodium seems to bear some relation to immunity. . . . Thus the Jensen sarooma when it recedes loses 40 per cent of its salts and at the same time the sodium content of the tumor increases 20 per oent, while the blood increases in salt content 37 per oent with a 19 per oent increaae in the sodium.” According to Rohdenburg, “Neoplasia is a oondition brought about by a disturbance of the salt metabolism of the oell, by reason of whioh the cell becomes hyperminerabed as oompared with the surrounding body juices.” He suggested that attempts be made to increase the sodium content of the blood stream. No independent evidence in support of such a theory has been found in the literature. 8wmmary: The literature contains no analyses on the amount of sodium per gram of dry tumor tissue. In persons suffering from various types of cancer, the concen- tration of sodium in the blood serum does not appear to differ from the normal value. The same holds true for strains of mice sus- ceptible or resistant to transplantable mammary adenocarcinoma. Sodium has neither an inhibitory nor an accelerating effect on the growth of transplanted tumors. Immersion of tumor tissue in solutions of sodium chloride prior to inoculation is without effect. Oral administration of sodium chloride is also without effect on either the number of takes or the rate of growth of the tumors. Pot assizcm Potmsium Costerst of Tzsmors: Beebe (19044) pointed out that rapidly growing human tumors were richer in potassium than degenerating tumors. He said: “The fresh, vigorous tissues, having active nutritive changes, contain a relatively large amount of the ion which has the power to call forth vital activity. There is no reason to suppose that the potassium ion by its presence has acted as the catalyzer to cause the rapid eell division in these neo- plasms; but its presence does seem to be in some way associated with their remarkable nutritional activities, of which rapid growth is one manifestation.” Examination of Beebe’s data reveals 9 analyses of malignant tissue, one analysis of a benign tumor, and one analysis ofa nor- mal tissue. Neither the potassium values alone nor the potas- sium : calcium ratios offer convincing evidence of any relationship between potassium content and malignancy. The scanty data can be considered only as suggestive. Soon after Beebe’s results appeared, Clowes and Frisbie (1905) commented as follows: “A few analyses of human tissues have been published by Beebe, who was, however, compelled to make we of material of various origins, which can scarcely be said to afford an accurate basis for cornparison. . . . Owing to the diffi- culty of comparing tumors of such diverse nature, generalizations as to age and potassium and calcium content are at least open to criticism.” Clowes and Frisbie analyzed 100 tumors; all of them were of the same type (Jensen’s mouse adenocarcinoma) and therefore offered “particularly favorable opportunities for such a comparison, being easily controlled and examined under varying oonditions of virulence, age and necrosis. ” In summarizing their extensive analyses of this particular mouse tumor, Clowes and Frisbie wrote: “It has been shown that rapidly growing, large tumors in mice mntah a high peroentage of potassium and little or no calcium; also that old, slowly growing, relatively necrotic tumors contain large amounts of calcium and little or no potassium. “It has been observed that in groups of comparable mice the potassium and calcium content is a function of the age and rapidity of development of the tumor. . . .” As was mentioned previously, the analytical data are not given in terms of dry tissue. The next important contribution was made sixteen years later by Waterman (1921), who pointed out “Qu’il htait dbirable de reprendre cette htude en la dhveloppant, ahde confirmer, si pos- sible, ces rhsultats admis sans critique dans la bibliographie. . . .” For example Neuberg (1911), in his review, stated that Beebe “hat die Mineralstoffe menschlicher Tumoren einer eingehenden Untersuchung unterzogen. ” As a matter of fact, Beebe’s study was far from being exhaustive, and Beebe himself had stated that hie experiments were too few in number to allow the drawing of definite conclusions. As regards the accurate determination of potassium in biologi- cal material Waterman (1921) said: “I1 est assez 6tonnant que pour la suite de la dbtermination des m6taux alcalins il ne semble paa encore exister de mhthode exclusivement applioable B cette d6termination et B l’abri de toute critique.” Employing the widely used ohloroplatinate method and the newly devised oobalti- 996 Me J. SHIOAB nitrite method of Kramer and Tisdall (1921), Waterman (1921, 1922) found, in agreement with the mnclusions of previous an- alysts, that the more rapid the growth of the tumor, the greater was its potassium content. His observations were made on a series of Rous fowl sarcoma, a series of miscellaneous human tumors, and a few other types of tumors. Rohdenburg and Krehbiel (1922, 1924) found that “the po- tassium, calcium, and sodium content of transplantable rat tumors vanes greatly, not only in the total amount of these mineral salts but also in the percentages which each forms of the total. In searching for the biological sijpificance of these variations, it wwi found upon analysis of four different types of rat sarcoma that the type containing the highest percentage of potaseium showed the greatest percentage of ‘takes ’ on inoculation. However, when tumors of markedly different origin and histological structure, such, for example, as the Flexner-Jobling rat carcinoma, and the Bullock and Curtis chondro-rhabdomyo-sarcoma,were analyzed, the results were different. The Flexner tumor has a muoh higher mineral content and 40.6 per cent more potassium than the mixed tumor, but when transplanted both growths show about the same peroentage of sumessful grafts. In connection with variations in the salt content of individual tumors, it was found that in a series of Sarcoma 10 and in another of Sarcoma 8 the tumor at- taining the largest size in a given period contained the largest amount of potassium. . . “0x1 the basis of these hdings it would seem that potassium has some significance in regard to the ability of the himor graft to survive in a new host and to its rate of growth.” Roffo (1924-25a) found that the potassium content of rat tumors was about twice that of the rest of the rat and waa about the same as the potassium content of rat embryo. He coneluded that these findings indicated that potassium plays an important part in proliferation of tissue. There is general agreement that an increase in the potassium content of tumors is associated with more rapid growth. It has been widely assumed that thie indicates a causal relationship, that an increase in potassium content results in greater proliferative wtivity. Before suoh a conclusion oan be mepted, however, one possible source of error should be ruled out. Since red blood corpuscles are rieh in potassium, and since rapidly growing neo- plasms have a good blood supply, inclusion of red corpuscles in the material . analyzed would inorease the potassium values. Therefore, contamination of the tumor tissue by red blood cells might conceivably account for reported high potassium values. SODIUM, POTL58CIIUM, OAbOIVllb AND XAQNESIUM IN OANOEB 997 Thie possibility could be verified, or ruled out, by simultaneous potassium and iron analyses. Potaiwiurrr Content of Blood: Because of the widespread belief that the potassium content of tissue is increased in cancer, num- erous efforts have been made to ascertain whether there is any correlation between blood potassium and cancer. Prior to the work of Waterman (1921) there was no reliable information on the potassium content of blood in cancer. Waterman analyzed blood eerum of normal and cancerous persons and concluded that, contrary to previous reports, the potassium content of blood serum is unchanged in camer. He emphasized the importance of avoid- ing even a slight cunonnt of hemolysis if correct values for serum potaesimn were to be obtained; since red cella are much riaher in potaesium than is blood serum, the occurrence of hemolysis results in erroneously high values for serum potassium. In ad- dition to taking precautions to avoid hemolysis, Waterman cen- trifuged his blood samples twicte in order to insure complete removal of red cells. In his careful analytical investigation he did not find any significant inorease in serum potassium in cancer. The next important-investigation was that of Theis and Bene- dict (1924). Them careful workers concluded that the potasaium content of serum was somewhat lower than nodin a large proportion of canoer patients; they considered a value of 20 mg. per 100 C.C. m the “low normal” value. Schepetinsky and Kdtin (1929) reported that the sem potassium was slightly below nodin about 40 per cent of their series of cancer patients; in the majority of the cases the values were normal. In a few cancer patients Ottonello (1926) obtained higher values for serum potassium than were obtained in normal in- dividuals. Louros and Gaessler (1929) and Gaessler (1930) ob- tained values dehitely higher than normal (see pages 648-9). The data of Blumgarten and Rohdenburg (1927) were obtained on whole blood and may be passed over without further comment. Fukunaga and Nagaoka (1928) determined serum potassium in rabbits before and after inoculation with a sarcoma. They noted a slight, but consistent, decrease in potassium as a result of tumor implantation. Pitts and Johnson (1931) found the average serum potassium for non-cancerous individuals to be 19.6 mg. per 100 C.C. and the average for cancer patients 21.2 mg. per 100 C.C. Umond and Cantegril (1930a) at first thought that they had found a slight elevation of serum potassium, but when they ex- amined a larger series of patients (1930b), they concluded that variations in serum potassium are rare and that when variations are obtained they are apparently independent of the cancerous 898 Bf. J. ‘SHIOas prooess itaelf. These authors (193Oc), however, continued their searoh for inoreased potassium values in cancer and reported find- ing it in the oorpuscles. They olaimed that the blood oella in oanoer oontained defhitely more potassium than normal. Their work, however, is not oonvincing. Cantegril (1931) recently re- ported additional ftndinge of a similar nature. Higher values for serum potassium were regularly obtained by Giacobbe (1931) in canoer we8 than in non-canceroue individuals. The oonflioting conolusions of various authors are due in part to laok of agreement as to what oonstitutes the range of normal variation of serum potassium. For this reaaon values which some analysts desoribe aa within the limits of normal variation are oonsidered by others to be pathologioal. Amording to Hawk and Bergeim (1931), “The potassium oonoentration of human blood serum is relatively constant, ranging from 16-22 mg. per 100 o.c.” Peters and Van Slyke (1931) give 19.5 mg. per 100 c.0. as the normal value. The best work indietitea that the average is about 20 mg. per 100 0.0. Effect of Treatmerct with Potassium: Oolddeher (1912) and hldzieher and Bosenthal (1913) found that subcutaneous adrnin- istration of potassium oitrate increased the weight of careinomas in treated mioe as compared with oontrol doe. Boneey and Wolf (1922), Troieier and Wolf (1922), and Wolf (1923) studied the effwt of immersion of mouse tumors in potassium ohloride solu- tions. They noted a shortening of the latent period in the po- taasium-treated tumors upon inwulating them into mioe ; the later development was not affwted. On the other hand, Sugiura and Benediot (1922) found that oral administration of potassium oarbonate resulted in a slight retardation of the growth of the Flexner-Jobling rat oaroinoma. Oral administration of potassium nitrate was without effect, Hiindel (1924) gave potassium ohloride by the oral route to mioe and inoculated them with a oaroinoma; he conoluded that administration of this salt resulted in an inoreaaed number of takes and in a stimulation of growth. However, the differenoe be- tween the group which received potassium ohloride and the control group was not great enough to be impressive. Roffo (19%25b, 1924-25c, 1925) oultivated tumor tissue irc vitro and found that growth was favored by the addition of po- tassium ohloride. In doefed a diet defloient in potassium, sarcoma 37 ceased to grow and frequently disappeared, amording to Langfeldt (1929). These mioe were immune against a seoond implant of this sarcoma after regression of the first implant ; they were not immune, how- ever, to subsequent implantation with Ehrlioh oaroinoma. 8icmmary: It is generally agreed that young, aotively growing BODIUM, POTASBIUM, OAWIUIYL bND MAGIPEBIUM IN CJANOBB 999 tnmore are richer in potassium than slowly growing or old tumors. The potassium content of blood serum is unchanged in cancer, amrding to most investigators. It has been concluded by various investigators that potaesium has a et’imnlating effect on tumor growth. Calciwn Calcium Content of Tumors: Beebe’s (19045) analyses in- dicated that rapidly growing, non-necrotic human tumors did not contain much calcium. In the extensive series of analyses of mouse carcinomas performed by Clowes and Frisbie (1905) it was found that small tumors had more calcium at a given age than large, rapidly growing tumors of the same age. Waterman (1921) analyzed tumors from various mammals and confirmed these fhd- hge: the older the tumor, and the slower its growth, the greater was its calcium content. With a series of Rous fowl sarcomas, analyzed at all stages of growth and degeneracy, Waterman similarly found that old tumors contained more calcium than fresh metastases in the same animals. Rohdenburg and Krehbiel (1922,1924) determined the calcium content of rat tumors but drew no conclusions regarding the r6le of aaldum. Roffo (192k25a) found no regular changes in the cal&um content of rat tumors of various ages. Goldfeder (1928) obeerved little difference between the calcium content of young and old tumors (mouse and chicken). She also found that there waa little difference between the calcium content of necrotic and partly necrotic areas of the same tumor. On the basis of analyses of a small number of rabbit sarcomas Nagaoka, Hirata and Fukunaga (1929) concluded that their re- sults corroborated the findings of previous workers as regards the high potassium: calcium values obtained with rapidly growing tnmors and the low values obtained with slowly growing or de- generating tumors. Eichholtz (1931) has recently repotted careful analyses of rat tumors. He found that, although the calcium content of tumors varied widely from rat to rat, yet the calcium content in different portions of the same tumor was remarkably constant. Further- more, different tumors in the same animal had the same calcium content. Rat and mouse tumors were analyzed for calcium and for the various forms of phosphorus by Buchwald (1930). When necrotic material was completely removed, the calcium and orthophosphate values were low; when necrotic areas could not be completely ex- cluded from the sample analyzed, these values were high. Buch- wald stated : The fairly good parallelism between orthophosphate and oalcium content makes it probable that necrotic areas contain considerable amounts of calcium phosphate. ” This oonolusion is in agreement with the opinions of earlier workers. Calcium Contest of Blood: In view of the inhibiting effect widely aeoribed to calcium, it waa expected that the calcium content of blood might be found to be diminished in oancer. However, the caloium level of both serum and plasma was found to be normal by Waterman (1921), Blum and Klotz (1923), Roe and Dyer (1930), and Pitts and Johnson (1930). It is true that Theie and Benediot (1924), Leicher (1925), Harnes (1929), Gunther and Greenburg 1930~)and de Fermo (1930) found that low values for serum oalcium are often obtained in cancer ;but it is the oonsensus of opinion at the present time that these low values are aeoribable to cachexia and to a physiological hypocaloemia of old age. It is generally agreed that there is no regular relationship between cal- cium content of serum (or plasma) and presence of malignancy. It is to be noted, however, that Fukunaga and Nagaoka (1928) reported that inoculation with Kato ’s sarcoma produced a definite increase in the serum calcium of rabbits. Effect of Calcium 0s Tumors ifi Viva: The development of tumors appears to be adversely affected by treatment with calcium. Goldzieher (1912) and Goldzieher and Rosenthal (1913) injected a solution of calcium lactate into tumor-bearing mice and obtained a deflnite reduction in the size of the tumors as compared with the controls, while Eatase, Kawamura and Mizutani (1920) reduced the number of successful takes of carcinomas and saruomas in mice by means of intraperitoneal inieotions of solutions of calcium salts. Hoshino (1922) found that calcium chloride interrupts the growth of transplanted rat sarcoma. On the other hand, Rohdenburg and Krehbiel (1922) found that calcium failed to cause any recession of mouse sarcoma 180. Sugiura and Benedict (1922) gave calcium ohloride by mouth to rats and found that it had a slight retarding effect on the growth of tumors. HZindel (1924) incorporated calcium chloride in the diet of mice and eoncluded that it reduced the number of takes and retarded the growth of the tumors. Uoldfeder (1928) gave a calcium chloride solution by mouth to mioe and to chickens ; tumor growth was delayed and a slight retarding effect was noted. Lmgfeldt (1929) found that a diet deficient in calcium had no in- fluence on the growth of a mouse sarcoma, and a slight retarding effeat on the prowth of two types of moum oaroinoma. Because of the numerous reports of the beneficial action of oal- cium on tumors, Shear (1933) tested the effect of calcium adminis- tered in a variety of wan. It was dissolved in the drinking water, and mixed with the food. It was injected subcutaneously, intra- peritoneallv. and intravenously. It was given alone, and in con- junction with various organic and inorganic substances. It was SODmM, POTASSIUM, 0-M AND MAONEWM IN OANOBB 1001 administered prior to and subsequent to inoculation with tumor theue. The calcium was given as chloride, lactate, citrate, and gluoonate. More than 1200 miw inoculated with sarcoma 180 or caroinoma 63 were employed. Since the results were essentially negative, the experiments were not applied to spontaneous tumors. Shear stated: “The few instances in which there was retardation of tu- mor growth, or recession of the tumors, may have occurred as a result of the poor condition of the mice and not because of any speeific aetion of calcium on the tumor. When the treatment is so drastic that the mice lose weight and vigor, caution must be exer- cised in assigning the cause of the mildly inhibiting effects that were sometimes noted. “The slight retardation of tumor growth which was occasion- ally noted in the treated mice was not regularly reproducible. If calcium does have a retarding effect on tumor growth, the effect is apparently only a minor one and is operative only when other factors are favorable.” Effect of Calcium on Tumors in Vitro: In addition to adminis- tering calcium salts orally and by injection to tumor-bearing ad- male, various investigators have studied the effect of treatment i~ oitro. Cramer (1918), in a careful investigation, immersed mouse tumors in solutions of calcium chloride and then inoculated them into mice. A definite inhibitory effect waa noted: the num- ber of positive takes was reduced, and the rate of growth of the tumors which took waa slowed down. That this was not due to impairment of vital function of the tumor cells was shown by the reversal of this inhibition when the calcium chloride-treated tumor tissue was immersed in sodium chloride solution; in such cases the number of takes was again 100 per cent and the rate of growth was increased. Cramer thus made the important 5nding that cal- oium diminishes the proliferative activity of tumors without de- stroying their vitality. Similar results were obtained by Sugiura, Noyes and Falk (1921) with a rat carcinoma. Immersion in solutions of calcium obloride, followed by inoculation into rats, showed that partial or total inhibition of ‘growth resulted, depending on the duration of the immersion. Treatment with calcium chloride which otherwise resulted in complete inhibition had no effect if the solution con- tained, also, potassium chloride and sodium ahloride. These sig- nificant experiments corroborate Cramer ’s important findings. Troisier and Wolf (1922), Rouesy and Wolf (1922), and Wolf (1923) also performed immersion experiments and found that irc vitro treatment with calcium solutions produced a slowing-down of the growth of mouse tumors. On the contrary, Kimura and Wada (1924) obtained no effect on tumor growth on treating four types of tumors with calcium solutions prior to inoculation. 1002 M. J. 8- Boffo (1924-256, 1924-25c, 1925) cultivated normal and malig- nant tiesue in uitro and found that mlcium hindered the growth of both, However, growth of tumor tissue was inhibited by amounts that did not prevent the growth of normal tissue; i.e. neoplastic tiasue is more sensitive to the inhibitory effect of calcium on growth than is normal tiesue. Erect of Parathyroid Hormorce : Gol&eher and Rosenthal (1913) reported that parathyroid extract had the same effect aa caloium lactate when injected into mice bearing adenoaarcinomaa, i.e. it retarded tumor growth. Slosse and Reding (1924), who obtained unfavorable results on treating cancer patients with oal- cium chloride, subsequently changed their opinion regarding the harmful effect of calcium and concluded that there is, in cancer, a decrease in the calcium ion concentration of the aerum which is due to a deficiency of parathyroid hormone. They (1929) there- fore treated patients with parathormone and recommended para- thyroid medication, not only as a form of treatment in cancer caaes, but also as a prophylactic measure. Yet Cherner and Elhsfiroff (1928), on injeoting parathormone into rats bearing Flemer carcinomaa, found that the tumors in the treated animals grew more rapidly than those in the controls. They were unable to obtain an inhibitory effect by means of parathyroid hormone treatment despite the increased calcium content of the serum and of the tumors. Similar experiments carried out with rat sarcoma 10 gave somewhat different results: no difference in the growth rate of this tumor waa produced. The number of animals em- ployed by Goerner and Shafiroff was apparently small: 9 rats bearing the ‘Flexner carcinoma were treated with parathormone and 4 were used ae mntrole; 6 rats bearing sarcoma 10 received parathormone and 6 were used as controls. The effect of parathormone was recently studied by Paik (1931), who also found that parathormone has a stimulating effeot on the growth of tumors in rats. Furthermore he found that calcium chloride, when injected into rats, produoed a deffnite diminution in the size of the tumors. The diametrically oppoeite effects of parathormone and of calcium chloride were rather puztiling. Paik tried to amount for this differenoe in behavior as follows : “I therefore postulate that difisible and non-diffusible organic calcium both stimulate tumor growth, but . . . there is an inhibitory effect due to inorganic calcium . . . it is noteworthy that the increme of blood or tiesue caloium ion produced indirectly by Parathormone is marked and muses a marked increase in tumor growth, while the inoreaee of calcium ion by calcium chloride solu- tion introduced directly in the blood or tissues generally inhibits tumor growth. ” Parathormone was also administered to tumor-bearing rats by Bischoff and Maxwell (1931). They say : “Hypercaloemia-pro- SODIUM, POTASSIUM, CUIUM AND MAQNEBIUM IN CANCEB 1003 duoing doses of parathyroid hormone administered daily to Hyde carcinoma rats produced no significant effect upon the rate of growth of the tumor or the mortality of the canoer-bearing ani- male as oompared with the controls. . . . Our experiment with the Hyde mrcinoma oonfirms the work of Goerner and Shafiroff on the Flexner and R-10 tumors: Parathormone has no inhibiting effect upon tumor growth.” In this experiment 17 animals were given parathormone and 22 rats were employed as controls. In other studies with transplantable rat saroomaa and car- oinomas Bischoff, Maxwell, and Ullmann (1931) found that “The experiments with parathyroid-thyroid-ectomized animals showed that these glands have no demonstrable effeet upon tumor growth.” Efect of Vitamirr D: The effect of vitamin D on the growth of Ehrlich’s mouse adenocarcinoma was investigated by Barelli (1929), who administered irradiated ergosterol by various routes to mice and concluded that this vitamin causes some depression of the growth of such transplanted tumors. Only 4 test mice and 4 oontrol mice were used in each experiment and apparently only 42 mice were used in all. Goerner (1930) inoculated 12 rats with Flexner rat carcinoma and administered irradiated ergosterol to 6 of them. The irradi- ated ergosterol had no effect on the rate of tumor growth or on the mloium content of the tumors. Caspari and Ottensooser (1930) employed a mu& larger num- ber of animals in studying the effeot of irradiated ergosterol on Ehrlich’s mouse carcinoma 5. They found that a rickets-produc- ing diet stimulated the growth of these tumors and that vitamin D in therapeutic doses only partially suppressed this stimulating effect. The effect of vitamin D was the same irrespective of whether it was given before or after inoculation, or during the en- tire experiment. The diet employed contained 3 per cent calcium carbonate and was of the type that produces a high-calcium, low phosphorus rickets. The finding that this diet stimulates tumor growth does not fit in with the widely accepted view that calcium exerts a retarding effect on tumor growth. Hosono and Narisawa (1931) added vitamin D to culture media in which chick embryo hearts were being grown. They reported that the growth of embryonic heart was better in the vitamin D- oontaining medium than in the control media. In plasma obtained from chickens tha.t had received vitamin D by the oral route, heart tissue grew more than twice as well as in control plasma. Cod-liver oil, irradiated olive oil, and irradiated ergosterol, with and without the addition of calcium lactate or calcium chlo- ride, were administered by Harde (1932) to female mice of a strain in which there occurred a high incidence of spontaneous mammary adenocarcinoma. In the preventive experiments, 3 groups of 8 mice were employed. An epidemic of paratyphoid carried off most 1004 M. J. S-AR of the test animals as well as the controls. Of the survivors (num- ber not stated) 3 treated mice showed mammary carcinoma. In the curative experiments Harde gave the same treatment, plus local applications to the tumors, to 30 mice; 2 cures were ob- tained. Since, in the course of six years, only one regression was observed among hundreds of spontaneous tumors, he concluded that vitamin A or D may be a useful adjuvant in treatment. The results obtained in these,experiments, however, cannot be regarded as demonstrating that the treatments had any beneficial action. Sumi (1930) fed cod-liver oil and irradiated ergosterol to rats and inoculated them with Fujinawa sarcoma. Control rats were given olive oil and non-irradiated ergosterol. A small number of rats were employed in each of five experiments. Sumi concluded that vitamin D exerts no perceptible influence on the growth of transplanted tumors. Budand Nakahara (1931) extracted the fatty oils from 4 kinds of transplantable tumors and found that these tumor oils contained vitamin D but not vitamin A. The tumors employed were: Roue chicken sarcoma, Fujinawa rat sarcoma, Flezmer-Jobling rat car- cinoma, and mouse carcinoma 63. The necrotic areas were re- moved before extraction. hkedCdcium: In recent years it has been alleged that the calcium ion concentration of blood serum (or plasma) is diminished in cancer. Waterman (1922), Rhmond, Sendrail and Lassalle (1925), Reding (1927), Reding and Slosse (1929), and Beregoff (1930) all concluded that the concentration of calcium ion in blood serum is below normal in cancer. It should be emphasized, how- ever, that none of these investigators obtained the calcium ion con- centration .by direct determination. The data on the amount of calcium ion present in serum was obtained by calculation from the formuJa of Rona and Takahashi (1913), which is

The authors substituted the experimentally determined values for the concentration of hydrogen ion and bicarbonate ion in this formula and, employing the value for potassium given by Rona and Takahashi, calculated the calcium ion concentration. Although this formula has been widely used, it does not give correct values for the calcium iorc coracentratiolz of serum. Ac- cording to modern theories of solution, in a saturated solution of caldum carbonate in equilibrium with solid calcium carbonate, the product of the ion activities is a constant, i.e.: am++ x Qcod' = K.P. where a,++ = activity of the calcium ion and awt = activity of the carbonate ion. SODIUM, POTASSIUM, OALOIUM AND MAGNESIUM IN CANOER 1005 When expressed in terms of ion concentrations instead of ion activities the solubility product constant, K‘a.p.,is given by [Cat’] X [CO,”] = K’..p. (2) where [Cat+] = total stoicheiometric concentration of calcium ions and [ CO,l’] = total stoicheiometric concentration of carbonate ions. The second dissociation constant of carbonic acid is given by

a& X amp = ki (3) %icO*J where a& - activity of the hydrogen ion amp = activity of the carbonate ion am-# = activity of the bicarbonate ion. In terms of ion concentrations this expression is given by [H’I X CC0,”I =$4 CHCO,’] (4) in which the quantities in the brackets are the total stoicheiometric ionic concentrations, instead of the ion activities. Combining equations (2) and (4), we get or

which iR the formula of Rona and Takahashi. It is thus obvious that this formula is based on the solubility product principle; collsequedy it is valid only for .colutiolls that are just saturated with calcium carbofiate. With solutions which are either undersaturated or supersaturated, the values of [Ca++] so obtained do not represent accurately the calcium ion concentra- tion. Although this formula has been widely used-for calculating [Caw] of serum, Rona and Takahashi did not use it for that pur- pose. In fact they stated: “AUSdiesen Versuchen kann gefolgert werden, dass, falls der geignete Bodenkorper fehlt, unter den im Serum vorhandenen Bedingungen der H’ und HCO, Konsentra- tion das Calciumhydrocarbonat ubersattigte Losungen von grosser Bestandigkeit su bilden befahigt ist. Auch direkte Versuche &en dies.” Furthermore their experiments showed that “In Beriihrung mit dem (entsprechenden) Bodenkorper ist also, wie 1006 M. J. SHEAB zu erwarten, die Loslichkeit bedeutend geringer, als wenn ein solcher fehlt. . . . Auf alle Fiille ist die Neignng an ubersiittigten Losungen eine sehr grosse, die bei den im Serum vorliegenden Verhaltnissen unter dem Schutze der Eiweisskorper noch erhoht werden wird. ” Rona and Takahashi did not emphasize that their formula is based on the solubility product prinoiple. They merely stated that they were using a formula whose validity was supported by the work of Bodllinder (1900). In deriving this formula, Bodhder stated that the expression Ca++X CO,” =R is valid for the equi- librium wher, cal&m carbonate is the solid phase. He then went on to combine this expression with the dissociation equilibria of carbonio acid. It is therefore clear that the calcium ion concentra- tion can be correctly calculated with this formula only in solutions in equilibrium with solid calcium oarbonate. In employing the formula of Rona and Takahashi for obtaining the calcium ion concentration of blood serum in cancerou~patients, Reding (1928) said: “L’exactitude de cette formule a 6t6 v6rif6e par Brook et Behrendt et sourtout par Brinckman et Van Dam en employant nne mhthode contr616e de dosage direct dn Ca ionid. Elle parast plus exact que la formule de Kugelmass et Shohl, qui tient compte en m8me temps de l’ion (HPO,).” Brock (1923), however, did not confirm the validity of this formula; on the con- trary, he recognized its weaknesses, for he stated: “Man muss zugeben, dass die Anwendung der Gleichung auf physiologische Verhaltnisse insofern eine Abstraktion bedeutet, als aie die etwaige Wirkung anderer Faktoren auf die Ionisation des Blutkalkes, wie 2.B. die Gegenwart der Anionen HPO,, die auf Grund klinischer Befunde neuerdings hervorgehoben wird, vernachliiesigt. Da jedoch gerade die Moglichkeit dieser Doppelwirkung unseres Wissens noch nicht theoretisch durchgearbeitet ist, wollen auch wir sie hier unberucksiohtigt lassen. ” Whatever be the merits of the method of Rrinkman and van Dam (1920), it was ueed by them only in ultrafiltrates and other protein-free solutions ; their work did not confirm the applioability of the formula of Rona and Takahashi to serum, as stated by Beding. On the contrary, Schulten (1925) found that the method of Brinkman axid van Dam produces supersaturated solutions and that this may lead to an indeterminable and perhaps very large error. Neither did the work of Behrendt (1924a, 19243) oonfer any such validity. Behrendt (1924a) used the method of Brink- man and van Dam on cerebrospinal fluid and concluded: “Es ergibt sich also die Tatsache, dass zur Erreichung des LGslichkeits- produktes mit Caw im Liquor die Bicarbonatmenge auf etwa das 24 fache erhiiht werden muss, die Phosphatmenge aber nur um 30 SODIUM, POTASSIUM, OALUIUX AND MAGWBEIUM IN OANOE~B1007 bis 50 Pros. Damit ist ein Esperimentalbeweis fur die grossere Bedeutung der Phosphate gegenuber dem Bicarbonat bedglich der Ionisation des Calcium8 in physiologischen Losungen erbracht. ” Subsequent investigations have borne out the opinions of Brock, of Behrendt, and of Kugelmass and Shohl (1923-24) regarding the importance of phosphate in calcium ion equilibria in physiological solutions. One of the reasons why the formula of Rona and Takahashi is not applicable to serum is that it entirely omits con- sideration of phosphate. Another serious weakness is the fact that, as far as calcium carbonate is concerned, such solutions do not behave as expected from simple solubility product considera- tions. Rona and Takahashi were well aware of this irregular and complex behavior of calcium and carbonate ions and cited previous investigations bearing on this point. Hastings, Murray and Sendroy (1927) employed modern the- ories of solution in their admirable investigation of the solubility of calcium carbonate in salt solutions and biological fluids. This important contribution showed that serum is not a simple system exactly saturated with respect to calcium carbonate, and that the equilibria involving calcium and carbonate ions are so complex in serum that simple generalizations cannot be drawn at the present time. Although equilibrium is readily obtained on shaking pro- tein-f ree solutions With solid calcium carbonate, such systems “when containing phosphate in solution, even in the presence of solid calcium carbonate, indicated an ppparent state of super- saturation with regard to that salt.’’ Many other curious anoma- lies were noted by Hastings, Murray and Sendroy. Sdce it to say that reliable information regarding the concentration of cal- cium ion in serum mnnot be obtained by the use of the formula of Rona and Takahashi. The amount of calcium ion in serum seems to be governed by the concentration of phosphate rather than of carbonate. The equilibria involving calcium and phosphate ions in serum have been studied in the light of modern concepts of solution by Holt, La Mer and Chown (1925), by Sendroy and Hastings (1927), and by Shear, Rramer and coworkers (1928-30). In the investigations of the two groups of investigators first mentioned attention was focussed on [PO,”’] and on the ion product [Cat+]*X [POI”’]’: ex- isting obscurities were not clarified and new paradoxes appeared. Shear and Kramer (1928) examined the calcium ion equilibria in- volving [HPO,”]; Shear, Washburn and Rramer (1929) showed that inorganic serum solutions came into equilibrium rapidly when shaken with crystalline dicalcium phosphate ; Shear and Kramer (1930) found that serum, too, rapidly attained equilibrium when shaken with dicalcium phosphate. Shear and Ofier (1931) stu- 1008 Y. J. SHE- died the binding of calcium ions by serum in the formation of un- ionized complexes and suggested the following formula : where [Ca++]- = concentration of ionized calcium in serum as drawn calmr rum = concentration of total calcium in serum as drawn [CaXIm, = concentration of bound calcium in serum at equilibrium. This principle had been used by Shear and Kramer (1930) in calculating the calcium ion concentration of serum from data ob- tained in equilibration experiments with dicalcium phosphate. Values of [Caw] calculated by this method should, however, be re- garded only as tentative approximations at the present time. Further study of the possibilities and limitations of the method is required before it can be known whether data on [Caw] obtained in this way are of value. In addition to the foregoing methods, [Ca++]has been estimated in blood fluids by dialysis and by ultrafiltration methods. Roffo and Correa (1925) determined the amount of ultrafilterable cal- cium in the plasma of cancer patients and found it to be about 62 per cent of the total calcium, i.e., about the same as for normal plasma. Gunther and Greenberg (1930a) determined the diffusi- ble calcium of the serum of cancer patients by ultrafiltration; they, too, found the concentration of diffusible calcium to be within nor- mal limits. . It is generally recognized that the diffusible calcium, i.e., the calcium which is obtained in the dialysate in dialysis experiments, or in the filtrate obtained by ultrafiltration, is not the same as ion- ized calcium. Such methods have been employed only for want of a better one. As Hastings, Murray and Sendroy stated, “In re- cent years, various methods have been employed in an effort to ac- complish a difficult task, the measurement of calcium ion concentra- tions in liquid systems in general, and in serum and biological fluids in particular. Many have attempted to obtain these values by the analysis of ultraflltrates, some by electrometric measurements with amalgam electrodes (Neuhausen and Marshall; 1922), some by colorimetric methods, and others by calculation from experiments with difficultly soluble salts (Rona and Takahashi, 1913), (Rugel- mass and Shohl, 1923-24), (Bolt and collaborators, 1925, a, b). The lattermethod is the one which . . . has been adopted for these studies. ” The numerous statements to the effect that in cancer there is a SODIUM, POTASSIUM, UALOIUM AND MAGNESIUM IN OANOEB 1009 &minution in the calcium ion concentration of serum (or plasma) have been based on calculations from the formula of Rona and Takahashi ;since this formula is inapplicable to serum (or plasma) these deductions are of no value. There exists no direct method for the accurate determination of the calcium ion concentration in suoh systems. Fosbinder (1929) made a careful study of the cal- dnm amalgam electrode and found “apparently insurmountable ditticnlties in using the calcium electrode for the determination of calcium-ion activities in physiological solutions. The potential is lowered by the presence of cations, either above or below calcium in the electromotive-force series. The potential is also lowered by the presence of protein.” Corten (1928-29) reported data on the caloium ion concentration of whole blood and serum, which were obtained by the third order calcium electrode of Corten and Ester- mann (1928). Le Blanc and Harnapp (1930), however, found that this electrode was unsuitable for the determination of calcium ion concentration in biological solutions such as blood. Of the indi- rect methods, none as yet has been shown to be capable of giving amurately the calcium ion concentration of serum. Miscellmaeozcs Btudie8: Calcium has been found to affect vari- ous physical and chemical properties of tumor tissue. Waterman (1922, 1923), in measurements of the electrical resistance and polarization of normal and tumor tissue, found that there was a difference in the behavior of the two types of tissue. Furthermore, treatment of normal tissue with calcium chloride shifted the P/R ratio in one direction while it produced a shift in the opposite di- rection in malignant tissue. Waterman (1928) subsequently ob- tained similar results with cell suspensions. X-ray treatment affwted this ratio in opposite ways in normal and cancer cells : the effect of x-rays was similar to that of calcium chloride. The investigations of Magath and his coworkers (1927-1930) on the absorption of water by tissues in vitro also showed that the effect of calcium on tumor tissue is different from its effect on normal tissue: the effect of calcium chloride on imbibition was diametrically opposite in the two types of tissue. The effect of x-ray treatment on imbibition was analogous to that of calcium ohloride. Finally, irradiation of tumors caused them to behave more like normal tissue when subsequently treated with calcium ohloride. tYummary: Many investigators have reported that calcium has an inhibitory influence on tumor growth. In tumors that are old or growing slowly, the calcium content is greater than in young, rapidly growing tumors. It is generally agreed that deposition of oalcium phosphate in necrotic areas probably accounts for the increased calcium in necrotic tumors. The question remains, how- ever, as to whether there is an increase in calcium in old tumora or in slow-growing ones apart from that due to pathological cal- cMcatioa No relationship has been established between the calcium con- tent of blood serum (or plasma) and cancer. Some observers have reported that the serum calcium of oancer patients is normal. Others have found values somewhat below normal, but have asoribed this hypocalcemia either to advanced age or to caohexia." A number of authors have concluded that in cancer there is a diminution in the calcium ion concentration of blood serum (or plasma). These conclusions have been drawn from data du- lated by means of the formula of Rona and Takahashi. Since this formula is inapplicable to such biologioal fluids, values of caldum ion ooncentration calculated in this way cannot be regarded as correct. Estimates based on ultrafiltration experiments showed that the diffusible calcium of serum is Within normal limits in oanoer. Administration of calcium salts to animals by various routes, it has frequently been reported, delays the appearance of trans- planted tumors and slows down their subsequent growth. This re- tarding effect was not obtained by Shear (1933) in an extensive series of experiments. Treatment of tumor tissue with calcium salts ifi vitro prior to inoculation into animals inhibits tumor powth. Culture of neoplastic tissue irc vitro is inhibited by cal- cium salts in concentrations that do not inhibit growth of normal tissue. Magrcesium Sweeping statements have been made repeatedly concerning the all-important part played in canoer by magnesium. These claims were not substantiated by adequate supporting evidence ; neither have they been borne out by subsequent investigatione. Magfiesium Cofiterct of Tumors: Analyses for magnesium were not made by Beebe (1904-5) because the amount of magnesium that he found in tumors was so small that quantitative determinations could not be carried out. The same state of affairs was reported by Clowes and Frisbie (1905). The work of Robin (1912, 1913a, 1913b, 1922) does not warrant further comment. Bolaill (1930a, 1930b) analyzed a few mouse tumors for magnesium. Eichholta (1931) determined the magnesium content of rat tumors employ- ing the new hydroxyquinoline method for the precipitation of mag- a Thh review covem the literature to the end of the year 1081. Attention im drawn, however, to the reriea of papen by Yorbvek (1088) who, aontrary to tbe experlenae of mort invatiqrton, found that the pohudum content of chicken waom WM rmollsr and the ddum content grmter than that of the musale in which the tnmor waa growing. He also found an Qorwrsd calcium eontent in the blood wrum of the tumor- bearing ehickena. 80DIUM, POTAWIUM, OALOIUY AND MAGNESIUM IN OANOEB 1011 nesium. He obtained relatively large amounts of magnesium, amounting to about one-third of the amount of calcium present in the 2 tumors analyaed. No other analyses of the magnesium con- tent of tumor tissue were found. The aocurate determination of small amounts of magnesium in the presence of comparatively large amounts of calcium and phos- phate is an analytical problem of some di5culty. It is not an un- common procedure to make the solution alkaline when precipitat- ing calaium oxalate, in the separation of calcium from magnesium. When phosphate also is present in the solution, some magnesium ammonium phosphate precipitates together with the aalcium ox- alate under these conditions, as pointed out, for example, by. Shohl (1922). Even when the contamination of the calcium preaipitate by magnesium does not produce a pronounced error in the calcium determination, it may cause a very large error in the magnesium determination in the analysis of materials which contain small amounts of magnesium. Thus in analyses of bone solutions, Kramer and Howland (1926) found that “&-slight error in the calcium determination produces an enormous error in the mag- nesium. ’’ Wsshburn and Shear (1932) studied the determination of cal- oium, magnesium, and phosphate in the presence of one another; they found that when calcium oxalate is precipitated at pH3, con- tamination by magnesium is avoided and correct values are ob- tainable for the magnesium in the calcium filtrate. Magnesium Content of Blood: Blum and Klotz (1923) were the first to determine the magnesium content of the plasma of cancer patients; the values obtained were the same as for normal plasma Theis and Benedict (1924) found that serum magnesium was be- low normal in half of their advanced cases of malignancy; the average values for both the early and advanced uases were, how- ever, within the normal range. The data of Blumgarten and Bohdenburg (1927) were obtained on whole blood. Louros and Gaessler (1929.) obtained values somewhat higher than normal in Mteen cancer patients. No other analyses of the magnesium con- tent of serum (or plasma) were found in the literature. Efect of Magnesizcm in Vivo; Since 1918 numerous papers have appeared making extravagant claims for the beneficial effect of magnesium salts in cancer. The findings of Regnault, of Dubard, of Beding and his coworkers, and of Delbet and his wworkers have been widely cited, but their conclusions regarding the favor- able action of magnesium in cancer have not been based upon con- vincing evidence. As a matter of fact, magnesium has been reported by a number of investigators to have a stimulatiwg effect on tumor growth. Itami (1918) obtained no effect in mice with 1012 M. J. SElUAE two types of carcinoma and obtained a stimulating effect with a third type of carcinoma. Katase, Kawamura and Mizutani (1920) obtained an increased number of takes of sarcoma in magnesium- treated rats. Hoshino (1922) noted an enhancement of growth of sarcomas in rats treated with magnesium chloride. Sugiura and Benedict (1922) found that magnesium salts had an accelerating effect on the growth of rat caroinoma. Kotzareff, de Morsier and Morin (1929) claimed to have con- bedDelbet 's findings, but their experimental data were meager ; curiously enough they, too, noted that the tumors in some of the magnesium-treated mice grew more rapidly than in the controls. Marnllaz (1929,1930) also claimed to have noted a retarding effect of magnesium on tumor growth, but his data are evm more scanty than those of Kotzareff and his associates. He, too, noted that under certain circumstances magnesium had a stimulating effect on tumors. Thomas and Kreitmann (1931) reported obtaining beneficial effects on thyroid epitheliomas of salmonoid fiah by the use of magnesium halides. Bola (1930a, 1930b), using a small number of mice, did not And that magnesium had any significant beneficial effect on tumors. MoDonald and Itami (1931) came to a similar conclusion. The alleged greater prevalence of cancer in some regions has been ascribed to a lower content of magnesium in the soil of those localities. According to Marchi (1930), in those parts of Italy where the salt contains comparatively large amounts of mag- nesium chloride the mortality from cancer is appreciably less than in those provinces where purer salt is used. Robi.net (19304 1930b, 1931) asserted that there was a striking parallelism between the death rate from cancer and the magnesium content of the soil, This was disputed by Gunsett (1929). 8chrumpf-Pierron f 1931a, 1931b) stated that malignancy is only one-tenth as common in Egypt as in Europe, and this is due to the comparatively greater magnesium content of the Egyptian soil. This statement was severely criticized by Brumpt (1931a, 1931b), who advanced data showing that cancer'is no rarer in Egypt than in Europe, the paucity of vital statistics in the former country being responsible for the supposedly low incidence. E3ffect of Magwsiwm I% Vitro: Roffo (1934-25b, 1924-26c) cul- tivated normal and neoplastic tissues in vitro and determined the effect of added potassium, calcium, and magnesium. He found that magnesium had an inhibitory effect on growth, but that the unfavorable action of magnesium was much less than that of calcium. Bummary: It is not known whether the magnesium content of tumor tissue is greater or less than that of corresponding normal BODIUX, POTASSIUM, O~XAND MAG~SIUM IN OANUBB 1013 tissue. The scanty available data indicate that magnesium is either normal or slightly below normal in cancerous blood serum (or plasma). The claims that magnesium salts have a beneficial effect in oancer have been made on the basis of clinical “impressions” and of experimental studies which were either poorly controlled or in which inadeqhate numbers of animals were employed. Other in- vestigators have reported that administration of magnesium salts to tumor-bearing animals is either without effect or has a stimulat- ing effect on tumor growth. Miscellarteozcs A number of authors have ascribed importance to the potas- sium: calcium ratio in cancer. It haa been stated that the effects of potassium and calcium on malignant growths are antagonistic; and that the greater the ratio of potassium to calcium the greater the malignancy. Embryo& Growth, Regeneration, etc.: Similarities have been pointed out between malignant and other growing tissues, such as developing embryos and regenerating tissues. Among recent con- tributions to this subject the following may be mentioned. Mangin (1930) found that the sodium and potassium contents of tbe chick embryo decrease with age and are considerably less in the fully developed chick than in the four-day embryo. While the reverse waa found for calcium, Mankin ascribed no importance to this, as it waa probably due for the most part, if not entirely, to bone formation. Kaufman and Laskowski (1931) investigated the potassium : oalcium ratio in growing tissues of normal pigeons. For the heart and for the brain the expected correlation between this ratio and growth rate was not obtained. The authors asoribed the increas- ing potassium content and decreasing calcium content to the in- crease in functional activity of these organs With age. For the eyes they found that this ratio falls regularly and steadily with the dearease in growth rate. Regenerating tissues of rabbits were analyzed by Bricker and Lazaris (1931). They obtained greater potassium : calcium values for these tissues than for normal control tissues obtained from analogous are- of the same animals. They stated that their re- sults paralleled those obtained by others on malignant tissue. The work of Roffo (192&25a), discussed on page 930, is also of interest in this connection. Miller (1923, 1926) found that rats failed to grow on a ration that contained less than 0.01 ,per cent potassium; they responded to the addition of potaesium ’in the diet or in the drinking water. 1014 M. J. SEUB Remy and Miiller (1931) added potassium salts to the diet of young rats and found that the increase in potassium had no ad- verse effect on growth, provided that the calcium: potassium ratio was kept at a satisfactory value (1:5) by the addition of calcium salts. The effeot of salts on the growth rate of paramecia was studied by P-rd (1926) in an attempt to correlate it with the findings of Bohdenburg et d regarding the changes in the mineral com- position of whole blood in cancer. Various concentrations of sodium, potassium, and calcium were employed in hay infusions. Packard found that when the hay medium had a sodium: caloium: potassium ratio similar to that of whole blood of rats with growing tumors, the paramecia were stimulated to divide at a faster rate than in the standard food medium. The absolute concwatrations of-the elements in the solutions were found to be less important than the balance between them. When the sodium: aabium ratio was disturbed, there was an immediate change in the division rate, a small increase in the sodium: calcium ratio resulting in an acceleration, while a large increase was followed by retardation. Inorganic ions were reported by Leupold (1928-29) as having the capacity of stimulating regenerative processes in wound re- p&r. Of the cations, sodium was found to be espeoially aotive. The experimeats were conducted on white mice with small incisqd wounds. This work was extended by Meyer-Dijrken (1929), who used a number of inorganic compounds. He concluded that mix- tures of inorganic salts have a growth-stimulating effect in en- dothelial regeneration, but that it could not be defbitely stated whether potassium is the chief agent in this stimulation. In a later study with various salts of sodium, potassium, and calcium, Meyer-Darken (1930), using improved methods, found no note- worthy effect. The results in the earlier, less exaot experiments, muld not, therefore, be unequivooally ascribed to the action of the salts used. Keiser (1925) analyzed tadpoles before and after metamor- phosis for calcium, magnesium, sodium, and potassium, and cor- related the data with growth. He utilized the ratio K+Mg/ Na + Ca in interpreting the growth intensity from a chemical point of view. The metamorphosis of tadpoles was also studied by Dodel and Jouve (1930), who immersed tadpoles in dilute salt solutions and observed the general growth and the various stages of metamor- phosis. They found that sodium chloride was without effect, that magnesium chloride retarded total and morphogenic growth, and that calcium chloride accelerated both types of growth. Plad Galls, etc.: Mention may also be made of some analogous studies on plant tiesues. Branhofer and Zellner (1920) reviewed BODIUM, POTASSIUM, OALUIUY AND MdONEBIUM IN OANOEB 1015 the smty literature on the inorganic composition of plant galls and gave analyses of plant galls and of normal plant tissue. They found that the calcium content of galls was lower and the potas- sium oontent higher than in normal plant tissue. These results present interesting analogies to those obtained on animal tumors ; however, only a few analyses were reported. Sorokin and Sommer (1939) investigated the effect of calcium and of magnesium on the changes in the cells and root tips of Phmacatiuum. Lack of calcium resulted in considerable damage. “The observed effects are attributed to some important part taken by Ca in the chemical constitution of the protoplast or because its absence so affects the physical condition of the colloidal system that normal mitoses are not possible.” James (1930) in a study of potato leaves found that senescence WM delayed by moderate additions of potash to the soil. It was saggested that addition of potassium ions increases catalytic ac- tivity in protoplasm, and that withdrawal of this ion is one cause of leaf ageing. James (1931) also analyzed the various parts. of the potato plant for potassium and found that for whole plants and for sprouts the relative growth rate paralleled the potassium content. Trelease and Trelease (1931) found that wheat seedlings grown in solution mltures showed characteriqtic injury when mag nesium waa present. This effect was decreased or suppressed by addition of calcium. Brooks (192930) discussed the potassium content of tumors in connection with his experiments on the ac- onmulation of ions in Vdonia. VIII. GENEBALSUMMABY It is obvious that unanimity regarding the r6le of sodium, potassium, calcium, and magnesium in cancer does not exist. The qhief reason for this state of affairs is that so few of the reported investigations are of a conclusive nature. The most that can be said of many of these studies is that the results obtajned are in- teresting and suggestive. In this aonnection the justifiably caustic comments of Woglom (1929) are so pertinent that they are quoted here at length. They are taken from the introductory remarks of hie excellent critical review op “Immunity to Transplantable Tumours,” in which the material reported in some 600 papers was reviewed. “Of these 600 oommunications a small number are of permanent value, many reoord unconscious and unnecessary repetitions of previous ex- periments, and far too large a proportion are merely ridiculous, for few investigators seem to realize that cancer research is a dimipline requiring some apprenticeship and that not everyone with an inoculating needle and a dozen white mice can plunge in and emerge with a discovery. . . . 1016 116. J. SHEAE “Furthermore, scarcely a week goes by which does not see de- duotions drawn from an experiment comprising six or eight mice, though the English school demonstrated years ago that hundreds are essential and Prime and others have more recently emphasized the same need. The Journal of Cancer Research, a periodical which those who undertake the study of malignant disease might reasonably be supposed to consult, has circillated article after article showing the extent to which tumours in untreated animals may vary in size, as well as the necessity of using a pure breed (Woglom, Bullock and Rohdenburg, Wood), but the flood of pub- lications continues as merrily as though neither neoplasm nor host contained a single variable factor. “It is not yet realized on all sides that extensive central necrosis represents a well-nigh constant feature of transplanted tmours, so that, as Caspari has espressed it, a medium-sized car- uinoma may actually contain less living tissue than a very small one; nor has the frequency of spontaneous absorption been thor- oughly grasped. Yet these elementary facts are aocessible to all. “The number of animals used is often omitted from the descrip- tion of an experiment, only percentages being given. Eiighty per cent seems imposing at first glance, for it might mean 80 mice out of 100; but on the other hand it may, and all too frequently does, mean 4 out of 5. An article is more valuable, too, when a writer does not withhold the species of animal employed in his work, or the variety of neoplasm, whether carcinoma or sarcoma. “Too many investigators regard the length of life following inoculation as a measure of the ‘virulence’ of a transplanted tumour, although a malignant neoplasm is, in itself, one of the most harmless things imaginable, exerting no evil effect until by ulmration it has become a portal of entry for bmteria, or until haemorrhage or metastasis has supervened. Therefore, an ani- mal with a small ulcerated tumour may die long before another with a large but intact growth. Furthermore, the lives of these small denizens of the laboratory are daily terminated by the vari- ous infectious diseases to which they are constantly exposed, with- out any reference to the presence or absence of a tumour. . . ’, “Finally, generalizations should not be based upon experiments with single tumour strains.” Although Woglom had in mind investigations dealins; with im- munity when he pointed‘out these common pitfalls in cancer re- search, his remarks apply equally well to studies on the inorganic chemistry of cancer. Systematic investigation of human tumors is obviously more difbult to carry out than studies with tumor-bearing experimental animals. As Clowes and Frisbie (1905) pointed out more than a quarter of a century ago, the clinical material used by some in- BODIUM, POTASSIUM, UAIAYIUM AND MAGNESIUM IN OANOEB 1017 vestigators is of such diverse nature that generalizations correlat- ing age with chemical composition are open to criticism. A simi- lar criticism was made by Roussy (1929) who stated: “Aux tra- vaux de MM. Slosse et Reding, comme 8. ceux de Vlas et de Coulon, de Waterman et de bien d’autres, on peut faire certaines objections. “Si l’on vent eviter toute cause d’erreur, il est indispensable, me semble-t-il,aqueles chimistes s’addressent, non pas au cancer en general, mais bien b une variCt6 bien pr&ise de cancer, variet6 de siege et de nature histologique bien d6finie. La localisation d’une tumeur sur tel ou tel organe, determine des troubles fort variables, suivant le cas, dont il faut tenir compte. “C’est pourquoi j’avais demand6 il y a quelques anndes 8. mes 6laves Peyre et Sanni6, de reprendre cette 6tude dans le sang et dans les tissue cancereux, mais en se limitant uniquement 8. une vari6t6 d’epithelioma: celle du col de l’ut6rus. cles recherches avaient abouti B des resultats qui ne sont pas tout 8. fait conformes B oeux qui viennent de nous &re prdsentds.” In investigations with experimental animals the complications resulting from a motley assortment of tumors are not encountered, for usually the same type of tumor is employed in any one study. However, different investigators often work with different tumors even when the same experimental animal is employed; this has given rise to contlicting reports regarding the effect of certain treatments on the growth of “tumors.” Furthermore, different species of animals are employed by different investigators in studying the effects of a given treatment or the distribution of a given constituent. This too has resulted in contradictory reports. One author, for example, will conclude that a given substance has an inhibiting effect on tumor growth because in 8 rats injected intravenously with this substance, transplanted sarcomas were smaller than in 5 control rats. A second investigator will conclude that this same substance has a stimulating effect on cancer because, in 10 mice which had received the substance orally, transplanted carcinomas grew, on an average, more rapidly than in 10 control mice. Still another will conclude that this same substance is with- out effect on malignant growth because subcutaneous administra- tion did not prevent the appearance of tumors in 6 tarred rabbits. Even when a single species of experimental animal is concerned and a single tumor strain is under consideration there are certain procedures which should be followed. In the first place, a suffi- ciently large number of test animals and control animals should be used in each experiment, and final conclusions should not be drawn until the experiment has been repeated again and again. Because of numerous factors, the same experiment repeated in an appar- ently identical fashion may give variable results. Positive value, therefore, can not be ascribed to a given material or treatment, un- 1018 Me J. SHBAB less beneficial results are regularly obtained in repeated experi- ments. So far as the mineral content of tumors is concerned, young, actively growing tumors appear to be richer in potassium than slowly growing or old tumors ; the reverse appears to be the case with calcium. Not enough data have beensobtained on the mag- nesium and sodium contents of tumors to warrant generalimtion. So far as the inorganic composition of the blood is concerned, no relationship has been established between the cancerous process and the amount of sodium, potassium, calcium, or magnesium pres- ent in the blood serum (or plasma). Here, too, the data are far from being complete. At the present time the situation may be summahd as follows : sodium appears to have neither an inhibiting nor an accelerating effect on tumor growth; potassium may have a stimulating effect, and calcium may have a very slight retarding effect ; magnesium does not appear to have the inhibitory effect that has repeatedly been claimed for it. None of these points has been established with sdcient car- tainty to warrant clinical application. Administration of para- thormone to cancer patients with the purpose of increasing the calcium ion concentration of the body fluids is obviously premature, to say the least, since no decrease in the calcium ion concentration of the blood serum or other body fluids has yet been demonstrated in cancer. Similarly, treatment of cancer patients with magnesium cannot be reoommended in spite of the favorable results reported by some clinicians. The analyses of tumors for magnesium are far too scanty to warrant any aonclusions ; the analyses of body fluids for magnesium have not established any relationship between mag- nesium and cancer; finally, administration of magnesium salte to experimental animals has failed to demonstrate a definite beneficial action and, according to some investigators, has indicated that magnesium may even have a stimulating effect on tumor growth. There is better agreement as regards the distribution of potas- sium and of calcium in tumors. Rut here, too, the results cannot be accepted without further substantiation : several possible sources of error must be ruled out before significance is to be at- tached to the increased potassium and decreased calcium values reported in rapidly growing tumors. Interesting leads have been uncovered which warrant further study, especially the reputed opposing effects of potassium and of calcium on tumor growth. Svstematic, comprehensive investiga- tions are required to clarify the present confused situation. SODIUM, POTABBIUY, OALUIUM AND MAQNESNM IN OANOER 1019 BIBLIOOWHY AD- K., AND ADLHI,M. : Strahlentherapie 42 : 684-90, 1931. BA~LLI,L : Ztsohr. f. 14rebsforsoh. 29 : 376-88,1929. BAUEE, E.: Ztsohr. f. Krehsfomh. 20: 368-74, 1923. B~BB~,8. P. : AIU. J. Phpbl. 12 : 167-72,1904-6. Bsmtmm, H.: Bioohem. Ztsohr. 144: 72-80, 19%. BBEBENDT, H.:Ibid. 146: 318-22, 1024b. BIEUEQO~,P.: J. Canoer Res. 14: 66487, 1930. BSBNADINI,L.: Indastria Chimioa 6: 410-17, 1930. BEBN~A., AND BEAYEL,J. J.: J. Bid. Cbem. 69: 113-24,1926. B-AUKI, E. : Ztsohr. f. klin. Med. 24: 460-611,1894. B~emom,F., AND MAXWELL, L. C.: J. Pharm. Exper. Therap. 42: 387-99, 1931. BIB~o~,F., MAXWELL, L. C., AND ULLMANN,H. J. : Soienoe 74: 16,1931. BIBEOP,W. B. 8. : J. Canoer h.Committee, Univ. of Sidney 1 : 34363,1930. BLUY,L., AND KLOTZ,A.: Compt. rend. Soo. de biol. 89: 1336-6, 1923. BL~CGABTEN,A. S., AND ROEDENBVBO,Q. L.: Aroh. Int. Med. 39: 372-84, 1927. BODLLVD~,Q.: Ztschr. f. physiol. Chem. 36: 23-32, 1900. Bo-, A. : Tmori 16: 201-11,193Oa. Born, 8.:Ibid. 16: 4204,1930b. Bow,A.: Bioohim. e terap. sper. 18: 77-91,1931. Boa~n,A., DE COULON,A,, AND Born, L.: Compt. rend. 800. de biol. 87: 1118-21, 1922. BBANHOF~EB,K,, AND ZELLNEB,J.: Ztsohr. f. physiol. Chem. 109: 166-76, 1920. BBIOKEB,F., AND hm8, J.: Ztsohr. f. Krehdorsoh. 54: 36-7, 1931. BBINXMAN,R., AND VAN DAM, E.: ha. Seot. Soi., Kon. akad. wetensoh. (Amster- dam) 22: 782, 1920. BBOCIK,J. : Bioohem. Ztsohr. 140 : 691-9, 1923. BEOOXE,8. C.: Protoplasma 8: 389-4l2,1920-30. Bmm, W. H.,AND PEAROE,L. : J. Exp. Med. 37: 601-29, and 631-46, 1923. BBUMPP,E. : Bull. Aoad. ma. de Paris 106 : 908-24,1931a. BBRXPT,E.: Bull. Aseoo. frang. p. l$$**+.pyg+?$: 443-67, 1931b. BBUXFT,E.: Bull. Aoad. ma. de P~#6!*2S405+19$$oi BU~ALD,K. W.: J. Clgg&&s, I$: 53d,' &!d:***" ** ~o~ngD. c. A., HUFF, !tx&&+$hqd,F., 1R. srqea~9e?aW, 1927. CIYEBON,Q. R.: J. Path: ah'd'Blrdt3d: BBe-5s,:lQkX:.. i..i:.:.-.*I:. CANTEQIUL, E.: Pm&e m6d. 68: 8O#,m,.$, A.M.@ Canoer Rev. 7: 2OO,1932. CASPABI, W., AND OTTEN~~O~EB,F.: j$t$tp?C$%wwh. 32: 7441,1930. CAmLm, R. : Lanoet 1 : 134,1907. .. . . CAW, F. : Canm Rev. 6: 401, 1031a. CA.PBIB, F.: Ibid. 6: 477-78, 1931b. CAVEBS,F.: Ibid. 6: 20-1, 1g31c. CAW, p. : Ibid. 7: 373, 1832. Cnr~rm,116. : Pathologioa 23 : 635-43,1931. Abst. in Chem. Abstr. 26 : 1976,1932, and Canoer Rev. 7: 276, 1932. C~~EK,A. J.: J. Pharm. Exp. Therap. 18: 423-47, 1921. Chmx, E. P., AND COLLIP, J. B.: J. Biol. Chem. 63: 461-4,1926. Cmm, 0.. H. A.: Kolloid Zeit. 16: 1234, 1914. Cmwm, Q. H. A.: J. Phya. Chem. 20: 407-60, 1916. Cmwq Q. H. A,: Proo. 800. Exp. Biol. Med. 16: 107-8,1917-18a. cIbwy9, Q. H. A. : Ibid. 16 :108-11, 1917-lab. Gmwaa, GI. H. A.: J. Canm Res. 4: 86-9,1919. Chowma, Q. H. A., AND FBIBBIE,W. 6. : Am. J. Physiol. 14: 17342,1006. Ckmrm, M. H.: Centralbl. f. allg. Path. u. path. Anat. 44: 1444,1928-20. Comm, M. H.,AND EB~XANN,I.: Ztsohr. f. phymol. Chem. 136: 228-30,1028. 1020 M. J. SHBAB

CUUEU, W.: Biochem. J. 12: 210-20, 1918. C~LE,O. W., TELKES,M., AND ROWLAND,A. F.: Am. J. Cancer 16: 2669-65, 1931. , DWET, P.: Presee ma. 36: 1473-7,1928. DELBET,P. : Bull. Assoo. franq. p. 1'6tude du cancer 19 : 176, 1930. DELBET,P., AND FRANICTEVIO': Bull. Aseoa. franc. p. I'btude du aaneer 19 : 742-61, 1930. DELBET,P., AND FUNIOEVIO: Bull. Aaad. ma. de Paris 104 : 668-76, 1930. DELBET, P., QODARD, H., AND PALIOS,C.: Bull. ha.franc. p. 1'Qtude du canaer 17: 516-24, 1928. DELBET,P., AND PALIOB,C. : Bull. Aeeoa. franc. p. 1'Qtude du cancer 17: 316-23, 1928~. DELBET,P., AND PALIOB,C. : Ibid. 17: 626-36, 1928b. DELBET,P., AND PALIOS, c.: Ibid. 20: 187-90, 1931a. DELBET,P., AND PALIOB,C.: Bull. Acad. mQd. de Paris 105: 50%9,193lb. DENIS, W.: J. Biol. Chem. 41: 3656, 1920. Do~m,P., AND JOm, P. : Compt. rend. 800. de biol. 104 : 1169-60, 1930. Dn- ': Bdl. Acad. mbd. de Paria 79 : 2964,1918. Dwaa~': Bull. Aasoo. frang. p. l'btude du canoer 8 : 148-68, 1919. DUB= ': Ibid. 9: 370-84, 1920. DUB^ ': Amh. d. mal. l'app. digeatif. 21 : 64867, 1931. Dumm, H.: Prem med. 39: 120-1, 1931. DU-, C. E. : Cancer Rev. 6 : 636, 1931. EIOHHOLTZ,F.: Bioohem. Ztsohr. 235: 170-3, 1931. EIOHHOLTZ,F., AND Bmo, R. : Bioahem. Zteohr., 226: 362-7, 1930. VON ETJLEE,H. : Soientia 60 : 209-18, 1931. DE FEUMO,C.: Minema med. 1: 8&7, 1930. FISKE,C. H.,AND SUBBABOW, Y.: J. Biol. Chem. 66: 376400,1926. FOBBINDER,R. J.: J. Am. Chem. Soa. 61: 1346-66, 1929. F~~END,H.: J. Biol. Chem. 61: 1164, 1922. FUKUNAQA,8, AND ~AGAOK+!&;.%~J&I& .J Path. goo. 18 : 244-66,1928. GAESSLEB,E. 0. : Stq$&w@iei3r$ @$it 1930. ~Aum~u'AND Hmd&s?':*hlf~ddnda.t'.'e~ syph. $04 964-6, 1899. OIACOBBE,C,;~i&~pym+-41'!;761+ .*3?.!;': ..a .::-:.* OOEENER, A. :*~..~~f.,~,i4;.~z;~OS~::**: 4 -!:' : OOHSNEB, A.;*%d ~~EAsIBo~.B,q,,P,p,;, J. Canoer Rae. 12: 294-300, 1928. OOLD~~8.: Zteahr. k&d$&+*~:603-36,1928. QOLDZWHEB : Verhandl. a.4htrcdhe pt~~%~~,usoh.16 : 2133-8, ign. OOLDZIEHER, M., AND hSENTRAL, E. : ztsahr. f. Krebeforeah. 13 : 321-31,1913. QUNS~T' : '' La fipartition du Cancer dane les trois d6psrtementa du Bas-Rhh, du Haut-Rhin et de la Moselle d'apda lee statietiques oftloielles 187411923. La Vie Sociale, Strasbonrg, 1929. (Cited by Robinet, L.: Bull. Bsaoa. franq. p. 1'8tude oanoer, 19: 243,1930. GUNTHEU, L., AND OBEENBEBO,D. M.: Arch. Int. Med. 46: 67-71,1930a. GUNTHER,L., AND OIUCENBRUU,D. M.: Ibid. 45: 983-1003,1930b. HALVIWION,J. O., AND BEBOEIM,0. : J. Biol. Chem. 32: 169-70,1917. HUBWEB, H. J.: Bioahem. Ztsahr. 71: 41643,1916. HXNDEL,M. : Ztaahr. f. Kxebsfomh. 21 : 281-7,1924. HARDE,E.: Compt. rend. Soc. de biol. 109: -7,1932. HAWEB,A. R. : J. Exp. Med. 60 : l0@-20,192Q. HABTINQB,A. B., MWIU~AY,C. D., AND SENDBOY,J., Js.: J. Biol. Chem. 71: 733-81, 1027. HAWK,P. B., AND BIE&OEIx, 0.:Practical Phyeiological Chemistry, P. Blakieton'e Son and Co., Philadelphia, 10th Ed., 1931. No iniW given. SODIUM, POTASSIUM, ClSMlIUM AND MAGNESIUM IN CANCER 1021 w,Q., AND EIUHHO~,F.: Bioohem. Ztsahr. 206: 2829,1020. H~Y,L. : J. Canoer Re%. 8 : 380-03,1024. BiBTtI, A. : Compt. rend. Soo. de biol. 88: 460-1, 1023. HO-N, T. : Bioohem. Zteahr. 243 : 146-0, 1031. HOLT,L. E., Ja., U &hit, V. K., AND CHOW”, H. B. : J. Biol. Chem. 64: 600-78, 1026. HOSHINO,T.: J. Osaka Med. 800. 21: No. 11,1022. Abet. in Jap. Med. World 3: 61, 1023. HOSONO,8., AND NABISAWA, 8.: Trans. Jap. Path. Soc. 21: 121-6,1031. ITAXI, S.: J. A. M. A. 72: 034-6,lOlO. JAMES,W. 0.: Ann. Bot. 44: 173-08, 1030. Abet. in Biol. Abetr. 6: 2238, 1031. JAMES,W. 0.: Ibid. 46: 42642,1031. Abst. in Chem. Abstr. 26: 167,1032. JENNEE,H. D., AND Kay, H. D.: J. Biol. Chem. 03: 733-48,1031. JINQUU,R.: Tm.J8p. Path. BOO.21: 743-7,1031. JONES,E. E., m RQmxg, M. D.: J. Canoer Ree. 11: 23245,1027. K~HN,H.: Klin. Wahneahr. 4: 178-9, 222-4,1926. KATAS~T.: Trans. Jap. Path. 600.12: 107,1022. KATASX,*KAWAXW,* AND &UTANI*: Verhandl. Jap. pathol. Gesellseh. 10: 107,1020. Cited by HLindel, M.: Ztsohr. f. Krebaforeoh. 21: 281-7,1924. mumm, L., AND LASKOWSKI,M.: Bioohem. Ztschr. 242: 424-36,1031. Kmsm, E.: Arah. exper. path. Pharmakol. 160: 663-86, 1031. Abet. in Chem. Abstr. 26: 618, 1032. ~EE,F.: Zeit. vergl. Phydol. 2: 463-73,1026. &mu, T., AND W~A,Y. : Qsnn 17 : 69-61,1023. gixmu, Y. : J. Bioohem. 8 : 460-04,1928. &am, B. : Bioahem. Zteahr. 236 : 614,1031a. Em, B.: Ibid. 237: 226-43,1031b. &UQ~ L: Bioahem. Ztaohr. 242: 226-32, 1031. KO~DSLBLBL,G,:Compt. rend. Soo. de biol. 02: 628, 1026. Kowog T. : Kyoto Igaku Zaesi 24 : 1024. Cited by Fukunaga and Nagaoh : Trans. Jep. Path. 600.18: 24436,1028. KONOW, P.: Z. mvrem. Chir. 3 : 831, 842,1028. Cited by MorBvek, V., Zteahr. f. Krebsfod. 36: 626,1032. K-, A., DE MOBSIEE,J., AND MORIN, A.: Bull. Aoad. de mu., Parie 102: 11-12,1g20. Kovh, L., AND AyBBoa, M.: Magyar orvosi aroh. 32: 220-6, 1031. Abet. in Chem. Abstr. 26: 4611,1031. Kmxm, B., AND OIWXAN, I.: J. Biol. Chem. 62: 363-60,lOM. -xu, B., m HOWLAND,J. : J. Biol. Chem. 68 : 711-10,1026. B., m TISDALL,F. F.: J. Biol. Chem. 46 : 33040,1021a. -xu, B., AND TISDALL, F. F. : J. Biol. Chem. 46: 467-73, 1021b. RR*yIEB. B., AND TISDALL,F. F.: Ibid. 47 : 476-61,1021c. ~PEHBIICL,0. : J. Canmr Rea. 6 : 100-203, 1920. bm,C.: Bioahem. Ztaohr. 161: 44043,1924. bmm, C.: Med. Welt, No. 40,1030. Cited by Adler, K,and Adler, M.: Strahlen- thempie 42: SWO, 1031. Kuamur~e,I. I?., m SHOHL,A. T.: J. BioL Chem. 68: 64s-S6,1924. Ism,F.: Compt. rend. Soe. de biol. 88: 1026-7, 1023. ham,E.: Skand. Aroh. Phydol. 66: 276, 1020. IUN~KI,8.: Zteahr. f. Krebsfod. 27: 116-24,1928. bmxr, A., m RQSENTHAL,0. : Bioohem. Ztschr. 207: 120-40,1020. Lo BLANO,M., AND -UP, 0.: Ztechr. Elektdem. 36 : ll6-17, 1030. Lmr- H. : Miinahen med. Wohnechr. 72 : 762,1925. Lmwrm, L., Bomm, Q., AND &m, E. : J. de pharm. et de ohiim. (8th Series) 6: 326-31, and 361-73,1027. No initialr given. lOn At. J. SHBAB LWY, P.: NBOplanmes 9: 1064,1930. Limpom, E.: Beitr. k path. Anat. u. t. allg. Path. 81 : 46-1OO,lea8-29. -8, M. R., m ANDIU~VONT,H. B.: Bull. Johns Hopldne Hosp. 42: 191-8,1928. Lmmmxi~,W.: Amh. Internat. Physiol. 16: 446-68,1920. Lrmems, D.: Ztsahr. f. Krebdorsah. 21: 3W6,1924. Lrmwm, J.: Bioohem. Ztsohr. 226: 33-1930. Lo- M., TUBPIN,R., AND ZUINn : Bdl. Aseoa. frnng. p. 1’8tude du mnm 14 : 322-7,1926. Lonmemrrm, F.: Clin. Ostet. 33: 129-39,1931. Abst. in Canoer Rev. 7: 881,1933. Lowoe, N., AND QAI~B~L~~:Zteahr. f. Krebsforsoh. 28 : 191-218,1929. Wommc, A. B. : J. Physiol. 32: 96-128,1906. MAQATX~,M. A.: Zteohr. f. Krebdomh. 24: 1264,1927a. MAQA~,M. A. : Ibid. 25 : 13236,1927b. MAWTX~,M. A., 4ND KOIX)YIJ=, M. J. : Zteahr. f. hbd~l.soh.30 : 467-72,1990. I(61Lle1~,J. : Ann. Burg. 93 : 180-9,1931. -KIN, W. El.: Med. J. Aaetralia 2: 41-8, 1930. ~K~KIN,W. R.: J. Cancer Res. Committee, Univ. of Sidney 3: 2324,1932. lhmr, C.: Riforma med. 46: 566,1930. ~U@U&,E. J. : NBOplaemee 9 : 321437,1930. MhaA, E. J.: Ibid. 10: 19744,1931. .:Bull. Bad. de ma. de Paris 102: 48-60,1929. lUtuxm5 : Ibid. 103: 166-72,1930. MODONAID, E. : Med. J. and Reoord 126 : 7967,1927. MODONUS, E. : Saienm 69 : 176-7,1929, MODONAID,E. : Ibid. 74 : 5660,1931. MODONAID,E., AND Illlurr, S.: Unpublished experiments, 1931. MEND-, P. : Compt. rend. Soo. de biol. 90 : 986-7, 1924. MEY~I-D~BXEN,Q.: Beitr. path. Anat. u. c. allg. Path. 83: 747-64, 1930. IV~YEU-D~BKE.N,Q. : Ibid. 86 : 866-64,1930. bfmmt, H. Q.: J. Biol. Chem. 66: 61-78,1923. Mxuut, H. Q. :Ibid. 70 : 687-91,1926. Moonu, B., AND WILSON,F. P.: Bioohem. J. 1: 297-327,1906. MO~VEK,V.: Ztsohr. f. Krebsforsoh. 36: 402-4504 609-22, 83640,1932; ibid. 36: 386-98, 68Q-36, 63746,1932. Momiux, J. C. : Aroh. Middleaex Hosp. 19 : 40-61 and 624,1910. M~EB: Zentralbl. f. inn. Med. No. 30,1909. Cited by Saheptinaky and KafStin : Arah. f. QyniLlr. 196: 37947,1929. NAGAOKA,T., HIBATA,M., AND Fuw~Aaa,S.:’Tnms. Jap. Path. Soo. 19: 667-62, 1929. NAKAYUB~,M, AND SU~UKI,K.: Trans. Jap. Path. Soa. 20: 677-9,1930. NiboBle, L. : Ann. Inet. de Pasteur 24: 12643,1910. Nhm, L.: Compt. rend. Soo. de biol., 86: 746-7, 1922. Nmm, C.: Zteohr. f. Krebaforsah. 10: 66-74,1911. NBUEAUBEN,B. S., AND MARSHATJ,,E. K,,Js.: J. Biol. Chem. 63: 96&72,1Saa. Ormmtto : NeOplanmea 10: 994,1931. Om-, P. : Polialinim (set. med.) 33 : 611-16, 1926. PAOKABD,C.: J. Caneer Res. 10: 1-14,1926. Pm, T. 8.: Tnma Jap. Path. Soo.20 : 853-4,1980. PAIK,T. 8.: Am. J. Canm 16: 276644,1931. Pm~um,R.: Riforma med. 41: 9374,1926. Pharm, H. :NBoplaemea 10 :16-26,1931. Pm,J. P., m VAN SLY- D. D.: Quantitative Clinical Chemistry, Vol. I., Williame and Willrins Co., Baltimore, 1931. No initials given. SODIUM, POTASSIUM, OUIUM AND MAGNESIUM IN OANOEB 1023

PI^, H. C., AND JOHNSON,H. R. : New England J. Med. 202: 416-23,1930. Pmore~,A., AND DOWBROW,8. : Ann. mat. path. med.-ahir. 1 : 163-71,1924. Po~~ore~,A,, AND Pmm, D.: Compt. rend. Soa. de biol. 92: 273-4,1925. DB FUADT,0. L. E.: Ztachr. f. Krebsforeah. 32: 698-8, 1930. RIQINQ,R.: Compt. rend. 800. de biol. 89: 817-8, 1923. REDINQ,R. : Amh. Internat. de m8d. explr. 3 : 612sO [712-801, 1927. R~INQ,R. : Canaer (Bnuellee) 5: 97-132,1928. REDINQ,R., AND DUSTIN,A. P. : Compt. rend. Soa. de biol. 88: 301-2,1923. REDING,R., AND SLOSSE,A. : Compt. rend. Soa. de biol. 88 : 644-6, 1023. REDINQ,R., AND S~~seg,A. : Bull. Aeeoo. frang. p. l’ltude du aanaer 18: 122-51, 1929. RJWNA~,J. : Bull. g6n. thbp. 168: 73740,1916-16. REGNAWLT,J.: Bull. Aaad. de mu. de Paris 80: 29-30,1918. R~YOND,A., AND CANNTW~BIL,E.: Compt. rend. 800. de biol. 103: 63-4, 1030a. RBYOND,A., AND C~BIL,E.: Ibid. 103: 764-6, 1930b. R~OND,A., AND CAN~OBIL,E.: Ibid. 104: 2934, 19300. RSMOND,A, SENDBAIL,M., AND LAEWAL~E: Compt. rend. Soc. de biol. 03 : 1061-3, 1936. ROYY,E., AND MULL=, A. : Klin. Wahnsohr. 10: 2338-41, 1931. RWAOD,A. : Bull. Aeeoa. frang. p. l’ltude aanaer 13: 288-97, 1924. ROBIN, A. : Bull. g6n. Wrap. 163 : 679-92, 1912. ROBIN, A. : Ibid. 165: 196-214, 1913a. ROBIN,A.: Compt. rend. had. de 80.166: 33443,19136. ROBIN,A. : Bull. doad. de m6d. de Paris 87 : 128-32, 1922. ROBINBT,L.: Bull. Aeeoa. frWQ. p. 1’6tude dn aanaer 19: 243-62, 1930~. ROBINET,L. : Bull. Ad. de m6d. de Paris 103 : 440-7,19306. ROBIN^, L.: Bull. Alesoo. frmp. p. 1’6tude du aanoer 20: 434-42, 1931. ROBINSON,E. P.: Med. Rtmrd 92: 36661, 1917. ROBINSON, E. P.: Ibid. 93: 7844,1918. ROBINSON, E. P. : Ibid. 98 : 1334,1020. ROE, J. H.,AND DYPI, H. M.: J. Cancer Res. 14: 3014,1930. J. H.,AND KAEN,B. 8.: J. Biol. Chem. 81: 1-8, lW29. ROm, A. H.:Bol. Inat. med. exp. 1: 307-23,1925a. ROm, A. H.: Ibid. 1: 493402,1926b. Roppo, A. H.:Ibid. 1: 53746,19260. ROm, A. H.: Nhplaemes 4 : 14866,1925. ROm, A. H.,AND CO~~~EA,L. M.: Bull. Soa. ahim. biol. 7: 522-6, 1025. ROm, A. H., AND ENOINA, 8.: Bol. Inst. med. exp. 1: 633-43, 1926. ROm, A. H.,AND LANDABUBW,J.: Bol. Inst. med. exp. 1: 83-02, 1025. Abst. in Phymol. Abetr. 10: 306,192626. ROHDIUNBUBQ,Q. L.: J. Cancer &es. 9: 3-10,1926. ROEDENBWO,Q. L., AND BZS~ARD,A.: Zteohr. f. Krebsfomh. 28: 301-10, 1929. ROED~NBUM,Q. L., AND KmHnxm, 0. F.: J. Cancer Res. 7: 417-37, 19‘32. ROED~NB~,Q. L., AND KREHBIEL,0. F.: Ibid. 8: 4334,1924. ROEDENBUBQ,Q. L., AND KREHBIEL,0. F.: Ibid. 9: 422-4, 1925. ~NA,P., AND TAKAHASHI,D. : Bioohem. Ztachr. 49: 370430,1913. C. 8. : J. Biol. Chem. 74: %7-83,1927. RO- M. D.: J. Biol. Chem. 78: 337-44, 1928. ROWSSY,Q. : Bull. ha.frang. p. 1’6tude du uanoer 18: 149-50,1929. RQWS~Y,Q., AND WOLF,M. : Arah. NBerl. Phyeiol. 7: 58270,1922. SOHAAL, Q., ~BWSOE~KAJA,M. A., AND ZWILICHOWSKAJA, E. J. : Strahlentherapie 40: 111-24,1931. ~OHICPETINSKY,A., AND &TIN, M.: &oh. Gyniik. 136: 379437,1929. No initials given. 1024 116. J. SHEAR

SCHLCBINGER,E. F.: Mitt. a. d. Qrencgeb. d. Med. u. Chir. 36: 47-64,19’& SCHBUMPF-PIEMION,P. : Bull. he.frang. p. I’gtude du eanoer 20 :307-22, 1931a. SOHBUMPPPIEBBON,P. : Bull. Acad. de ma. de Paris 106 : 818-23, 1931b. SOHIJYPEN,H. : Biochem. Ztaohr. 164 : 4762,1926. Sm,0. H., AND HOBNINO,E. S.: Am. J. Path. 8: 329-32,1932. SENDBOY,J., Js., AND HAHTINQB,A. B.: J. Biol. Chem. 71: 783-1348, 1927. SHEAR,M. J.: U. S. Pub, Health Rap. 48 : (in press), 1933. SHEAE,M. J., AND ~EB,B.: J. Biol. Chem. 79: 126-46, 1928. SHBUB,M. J., AND KRAM~B. : Ibid. 86 : 67744,1930. SHB~~B,M. J., WABEBUBN,M., AND KBAMq B.: Ibid. 83: 697-730,1928. SHEAB,M. J., AND OFFNEB, M. M.: Ibid. 91: 291-306, 1931. SHOEL,,A. T.: J. Biol. Chem. 60: 627-36,1922. Smsiu, A., AND REDINO,R.: Bull. Acad. Roy. m6d. Belg. [6] 4: 268-90,1934. Stossq A., AND REDINO,R. : Ibid. [6] 6: 166-73,1926. ~OBOKIN, H., AND SOMMBBL,A. L.: Am. J. Bat. 16: 23-39, 1929. Abet. in Biol. Abstr. 6: 2239,1931. S~KLO~A,J.: Deutsohe med. Wchnschr. 61: 106740,1926. SUQIUIUL,K.,AND BENEDICT,6. R.: J. Canaer Bee. 7: 329-69,19!22. SUOIWBA,K., NOYES, H. M., AND FALK,K Q.: J. Cancer Res. 6: 286-303, lV3l. SUMI,M.: Qann 24: 23040,1930. SUMI,M., AND NAKAHARA, W. : Sai. Papem Inst. Phys. Chem. Bee. 16: 69-74,1031. SVPLHW,K. : Compt. rend. Soo. de biol. M: 980-1,1926. Tms, R. C., AND BENEDI~,8. R.: J. Canoer k.8: 499403,1924. THOMAS,L., AND KBR~~MANN,L.: Bull. dead. de mU., Parie 106.: 1244,1931. TISDW, F. F.: J. Biol. Chem. 66: 43M,1923. TOKUNAGA,M., AND NAQAO~T.: Trans. Jap. Path. Soa. 19: 8624,1929. TM~,M. : J. Med. Ree. 14: 447-64,1006-6. Tna.r.~.*n~,8. F., AND TBELEASE,H. M.: Bull. Torrey Bot. Club 68: 127-48, 1931. Abst. in Cham. Abstr. 26: 168,1932. TBOIBIEE,J., AND WOLF,M. : Compt. rend. 800. de biol. 86 : 661-2, 1022. UEI,K. : Folia japon. pharmaool. 2: 288-94,1926. Abet. in Ber. Physiol. 37: 308, 1926. VOOBEOEWE,N. : Biochem. Ztaahr. 30 : 196-206, 1911. VOO~OEVE,N. : Ibid. 32: 304409, 1911. DE WAARD,D. J. : Biochem. Ztaahr. 97: 18643,1919. WARBUIW,0.: The Metabolism of Tumom, transl. by F. Diokene, Riahard R. Smith, Inc., New York, 1931. WASHBUBN,M. L., AND SHEAB,M. J.: J, Biol. Chem. 90: 214, 1932. WA-AN, N. : Arch. NBerl. Phyaiol. 6: 306-27,1921. WA~MAN,N. : Bioohem. Ztaahr. 133 : 636-97,1922. WATEBMAN,N. : Compt. rend. 800. de biol. 89 : 818-21, 1923. WARMAN, N.: Zteohr. f. Kmbsforsoh. 20: 376-93, 1923. WATERMAN,N. : Ibid. 27 : 2’28-40, 1928. WOOLOY,W. H.:Canoer Rev. 4: 129-214,1929. WOLF,M. : Compt. rend. Aoad. de 80.176: 10324,1923. ZADIK, P.: Ztsahr. f. Krebaforsch. 30: 473-81, 1930. ZIYMERN, A., AND WIOREAM,Y. L.: Bull. Aeeo~.franc. p. 1’6tude du cancer 17: 3824,1928. ZONDEK, 6. Q.: Biochem. Ztaahr. 121: 76-86,1921, ZWAABDEMAKER, H. : Am. J. Phyaiol. 45 : 14746, 1918. ZWAABDEMAKER,H. : J. Phyaiol. 63 : 273-89,1920. ZWAABDEMAK~H.: Arch. NBerl. Phyeiol. 4: 17746,1020. ZWAARDEMAKEU,H.: Klin. Wchneohr. 1: 636-7, and 6044,1922. No initialr given.