Cancer of the stomach : a clinical study

Chapter I. General Etiology

Cancer in General— Question of its Increase— Has Cancer of the Stomach Increased in Frequency ? — Statistics of Cancer of the Stomach at the Johns Hopkins Hospital — Sex— Age — Eace — Conjugal Condition — Family History of Cancer — Tuberculosis— Previous Stomach Troubles — Alcohol— Miscellaneous Conditions— Trauma — General Conclusions . .

Chapter II. Cancer Of The Stomach In The Young

Introduction — Cancer of the Stomach in Childhood; Analysis of the Recorded Cases — Cancer of the Stomach in the Second Decade ; Analysis of the Recorded Cases— Cancer of the Stomach in the Third Decade ; Report of Cases- Peculiarities of Cancer of the Stomach in the Young . . 16

CHAPTER IX STUDY OF THE TUMOUR (continued)

Tumours of the Pyloric Region — Some Common Features— Three Groups.— I. With Thickening and Induration— Cases— Estimation of the Degree of Involvement— Diagnosis from Thickening about a Chronic Simple Ulcer and from Simple Hypertrophic Stenosis of the Pylorus— Cases of the Former. —11. Tubular and Small Nodular Tumours— Symptoms— Non-malignant Tumours of this Region — Involvement of the Pylorus in Gall-bladder Adhesions. — III. Large Nodular Growths 92

CHAPTER X. STUDY OF THE TUMOUR (continued)

Tumours of the Stomach Wall— General Features— Tumours of Anterior Wall— Tumours of Posterior Wall— Massive Tumours of Wall — Tumours found at Autopsy, but not recognised during Life , 104

CONTENTS

CHAPTER XL

THE BLOOD IN CANCER OF THE STOMACH.

PAGE

Red Blood-corpuscles: Counts in Fifty-nine Cases— Haemoglobin : Estimations in Fifty-two Cases—Leucocytes, with Differential Counts— Digestion Leucocytosis— Cancer of the Stomach simulating Pernicious Anaemia li:

Chapter XII. Course Of The Disease— Latent Cancer

The Statistics of Duration of Sixty-three Cases— Conditions under which Improvement Occurs — Acute Cases — Chronic Cases— Latent Carcinoma of the Stomach— Three Groups of Cases : {a) With no Local Features, (b) With the Symptoms of an Associated Lesion, (c) With Metastases which mask the Trimary Disease 125

Chapter XIII. Pa1hological Anatomy

Position of the Growth— Presence of Adhesions— Ulceration — Size of the Stomach — Dilatation — Contraction — Metastases — Perforation— Summary of the Cases— Peritonitis— Ascites— Jaundice— Tuberculosis— State of the Lungs— Circulatory System — Miscellaneous — Classification of Gastric Cancers- Secondary Cancer of the Stomach 138

Chapter XIV. Treatment

Importance of Early Diagnosis and Exploratory Operation — Medical Treatment— Diet, Stomachics, Lavage— Special Symptoms— Pain, Vomiting, Constipation — Surgical Treatment: Exploration, Eadical Operation, Palliative Operations, Gastrostomy for Cancer of the Cardia, Gastro-enterostomy for Stenosis of the Pylorus , 148

Chapter I. General Etiology

CANCER IN GENERAL — QUESTION OP ITS INCREASE HAS CANCER OP THE

STOMACH INCREASED IN FREQUENCY? STATISTICS OF CANCER OF

THE STOMACH AT THE JOHNS HOPKINS HOSPITAL SEX — AGE RACE

CONJUGAL CONDITION FAMILY HISTORY OF CANCER — TUBERCULOSIS

PREVIOUS STOMACH TROUBLES — ALCOHOL MISCELLANEOUS CONDITIONS

TRAUMA GENERAL CONCLUSIONS.

In presenting the subject of cancer of the stomach in some detail we feel that no apology is needed. Statistics vary as to which organ of the body holds the unenviable position as the one most often attacked by malignant disease ; but however this may be, of internal cancer, not accessible to direct observation, that of the stomach is by far the most frequent and important.

While there has been a most gratifying decrease in the death-rate from the infectious diseases, sanitarians and statisticians have of late called attention to a marked increase in the mortality from malignant disease. Fortunately the question has aroused widespread interest, and even governments (State of New York) have been induced to make large appropriations for the study of the subject.

A glance through any recent volume of the Index Medicus, under the heading of Cancer, gives an idea of the enormous activity of different workers in all parts of the world. Pathologists and bacteriologists are seeking to discover the cause of the disease ; clinical physicians are perfecting in every possible way the means of diagnosis, so as to secure a very early recognition ; while the surgeons are daily widening the field of operative interference, and improving the technique of the older operations.

Since the organisation of the Johns Hopkins Hospital there is no single affection which has received more careful study than cancer. The article by Professor Welch on " Cancer of the Stomach " in Pepper's System of Medicine, the most exhaustive in any language, appeared when the pathological laboratory was the only part of the hospital plant in working order. We may refer to the work of Flexner and of Gilchrist on the so-called cancer organisms, to the introduction by Halsted of an ideal operation for breast cancer, with which most remarkable results have been obtained, and in the gynecological department to many observations on malignant disease of the uterus, on which subject Cullen, one of Professor Kelly's assistants, has published an elaborate monograph.

In the medical wards, where malignant disease in general is of relatively less importance, the stomach — the internal organ which perhaps takes first rank in frequency of attack — comes within our province. As will be mentioned later, our experience is unusually large for the size of the service, which is explained by the fact that the institution is a consultation hospital for a very wide area of country. We propose in this work to give the experience of the first nine years of the existence of the hospital.

The Increase of Cancer in General. — Before taking up the special subject in hand, it may be of interest to consider briefly the incidence of malignant disease in general. If figures can be trusted, the death-rate from cancer has greatly increased in all countries in which careful returns are made. Bryant,1 who discusses the whole question in the Wesley M. Carpenter Lecture, 1894, concludes that the death-rate from malignant disease is constantly increasing. Roger Williams2 estimates that the proportionate cancer mortality is more than four times greater than it was fifty years ago.

The census figures of the United States for the last five decades are as follows : —

Year.

Population.

Total Deaths.

Deaths from Cancer.

Cancer Deaths per 1,000 Deaths.

Cancer Deaths

Living.

875,521

2,088

3,672

6,224

13,068

18,536

9-3

12-6

17-2

21-1

9-0 11-7 16-1 26-0 296

1 New York Med. Jour.,Maj 18th, 1895.

2 Med. Press and Circ., London, 1896, N.S., LXIL, 232.

la the returns for 1880 there were 1,781 cases in which the cause of death was given as " tumour." If these be added to the above, the figure per 1,000 deaths is 19*6, and per 100,000 living 29*6. In 1890 there were 2,448 cases, which, if added, give figures of 23-9 per 1,000 deaths, and of 33*5 per 100,000 living.

In the Report of the Registrar-General for England and Wales for 1896 the cancer figures are arranged in groups of five years, from 1861 to 1895 inclusive. The death-rate from cancer per 100,000 living was as follows : —

Years . . .

1861-65

'66-70

71-75

'76-80

'81-85

'86-90

'91-95

Death-rate .

36-78

40-38

44-56

49-36

54-76

63-16

71-22

76-4

78-7

In 1896 the death per 100,000 living was 76-4, in 1897 78-7. These figures show a constant increase during a period in which the death-rate from all causes has steadily fallen. The total death-rate was in 1861-65 per 1000 living 22*58, and in 1891-95 it was 18*73. The intervening periods show a steady decrease.

Our colleagues in Australasia have demonstrated the same thing. The mortality figures for Victoria1 for the years 1870-84 show an increase at about the same rate as in England. Mnllins2 finds an increase from 1857 to 1893 in certain areas. MacDonald 3 points out the steady increase in the cancer mortality of New Zealand.

In the Registrar-General's Report for England and Wales for 1889, in discussing this question of increase, it is pointed out that the death-rate from cancer has risen more rapidly in males than in females, and that the increase has been greater and greater the more advanced the age. In 1851-60 the death-rate per 100,000 living was 19*5 for males and 43*4 for females; while in 1871-80 the rate was 31-5 for males and 62*2 for females. The increase was nearly general in all the organs affected. In the search for an explanation, greater accuracy in diagnosis is not considered adequate to account for the increase. But King and Newsholme,4 in an article " On the Alleged Increase of Cancer,1'

1 Austr. Med. Jour., Melbourne, 1887: N.S., IX., 99-101.

consider that the apparent increase is largely due to more accurate diagnosis. They endeavour to prove that the increase is only in cancer of organs which are " inaccessible." The rate in accessible organs, in which the disease is easy to diagnose, they consider to have remained about the same. They claim that the statistics of Frankfort-on-the-Main, where records are kept as to the organs affected (from 1860 to 1891), show an absence of any increase in cancer in these parts. W. Roger Williams combats this view strongly, and sums up his conclusions in the following words : " . . . the uniformity in the variations of the increments of increase in the long accession of years ; the fact that the increase has not been confined to one or a few parts of the body, bnt has involved them all — on the whole without any considerable disturbance of the normal proportionate ratios ; the fact that the increase has been diffused over the whole country, instead of being limited to certain areas ; and the fact that similar increase has taken place in most civilised communities where statistical records have been kept. These considerations, together with other facts, point conclusively to the reality of the increase in cancer . . . ." Welch1 points out that the increase may be due to the decreased mortality in infants, and the prolongation of life by improved sanitary conditions. He also quotes the statement of Dunn 2 that the cancer rate of a country may be taken as an index of its healthfulness. Payne3 combats the idea that the increase of cancer is only apparent, and holds that the explanation of improved diagnosis of cancer in the inaccessible organs does not account for it.

The interesting question of the geographical distribution of cancer is too large to be discussed here. The census returns of the United States for 1890 contain statistics and maps which bring out strikingly the variations of the cancer rate. Among these is the marked difference between certain areas in the northern and the southern states. Thus the figures for certain coast regions for 1890, per 100,000 living, are : For the South Atlantic coast region, 12 ; Middle Atlantic ditto, 21-9 ; and North Atlantic ditto, 34-7. In England marked geographical differences have been pointed out, and

1 " Cancer of the Stomach," Pepper's System of Medicine, Vol. II.

2 " Causes of the Increase in Cancer," Brit. Med. Jour., 1883, I.

3 Lancet, Sept. 16th, 1899.

Haviland says : " The cancer-fields of England are to be found in the sheltered and low-lying vales traversed by fully formed rivers, and having sites composed of the more recent geological formations."

Has cancer of the stomach increased in frequency? — To this question we find difficulty in presenting any positive answer from returns, as in a majority of these the cases of malignant disease are grouped together without any distinction as to the organ affected ; and, frequently, the diagnosis of cancer of the stomach is of doubtful correctness. In the United States census the returns for cancer of special organs are only for certain areas, and so cannot be taken as representing an exact general conclusion. In the census returns for 1880, in certain groups, among 1000 deaths from cancer where the seat of the disease was known there were 30018 from cancer of the stomach. The total deaths from this cause were 2133, which, in a population for the area of about 29,000,000, gives one death to 13,595 living. Certain areas in 1890 gave a rate of one death to 9,761 living.

Comparing the death-rate from this cause to the total deaths, Welch1 estimates that in New York from 1868 to 1882 the ratio was about 0*4 per cent. If only the deaths over twenty years be taken the ratio is 0'93 per cent. In the census returns for 1890 in certain areas in which the stomach cancer death rate is given, its ratio to total deaths is 0:49, and to deaths over fifteen years is 0*87. If, however, one-third of cases of cancer of the liver 2 be added, the ratio is 0-56 to the total deaths, and- TO 1 to deaths over fifteen years.

In the Report of the Registrar-General for England for 1889 figures are given for certain areas. The years 1868 and 1888 are taken for comparison. All cases of cancer of the generative and mammary organs in the female are excluded, and the death-rate is calculated for the population of thirty-five years and over. The death-rate per million living, of thirty-five years of age and upwards, from cancer of the stomach was in 1868 — males, 283-65 ; females, 193-45. In 1888 it was— males, 346-15 ; females, 277-75. This is an increase in males of 22 per cent., and in females of 44 per cent.

1 Loc. cit.

2 Welch, in arriving at his conclusions, added one-third of the cases of cancer of the liver as secondary to gastric cancer.

The relative proportions of cancer of various organs should give some information. If, as King and Newsholme 1 argue, the increase is only apparent, and due to better diagnosis of cancer in the so-called " inaccessible " organs — of which the stomach is a good example — the relative death-rate of gastric cancer should show a marked increase in comparison with cancer of the uterus — one of the " accessible " organs (uterus, breast and tongue) which they consider is rarely overlooked. But this not the case : in fact, as the Registrar-General's report in 1889 shows, the increase of gastric cancer is rather less than that of cancer generally. The figures for the United States for the known areas in 1880 and 1890 are given in the following table. The 1880 figures are from 7,111 cases, and the 1890 from 9,657. With them are also given the figures arrived at by Tauchon 2 from 9,118 cases. The figures represent the number of cases among 1000 of all cancers : —

Organ.

Tauchon.

Stomach .

Uterus .

Breast Liver

300-18 331-88 151-07

71-77

278-10

312-46

328-58

252-58

217-26

63-39

These rather scanty data do not justify a positive conclusion as to the increase of cancer of the stomach.

Cancer of the Stomach at the Johns Hopkins Hospital.—

From its opening, May 5th, 1889, until March 31st, 1898, 150 cases were diagnosed as primary cancer of the stomach ; of these 2 were cases of multiple primary cancer. During the same period there were 3 cases of secondary cancer of the stomach. Among the first 1000 autopsies in the hospital there were 40 of primary gastric cancer. The total number of admissions to the medical wards during this period was 8464. This gives a ratio of 1 case of gastric cancer to 56-4 admissions. Dr. Fitz has very kindly given us the figures for the Massachusetts General Hospital for the years 1889 to 1897 inclusive. Of 11,812 admissions in the medical wards there were 129 cases of cancer

1 Loc. ait.

? " Eecherches sur la Frequence du Cancer," Gas. des Hoj)., Paris, 1843, p. 313.

of the stomach — a ratio of 1 to 9T5. Dr. Wyatt Johnston has sent the figures for the Montreal General Hospital for a corresponding period. Among 9458 cases there were 54 of cancer of the stomach, a ratio of 1 to 175. These figures show a remarkable difference in three hospitals occupying almost identical positions in their respective communities. The figures of two of the large London hospitals may be given for comparison. At St. Thomas' Hospital during the ten years 1887 to 1896 there were 98 cases of gastric cancer among 18,958 medical admissions — a ratio of 1 to 193"6. At St. Bartholomew's for the same time there were 106 cases among 23,500 medical admissions — a ratio of 1 to 221*6. A certain number of cases of diseases of women were, however, treated in the medical wards. The large number of cases in our series is due in great part to the fact that this is a consultation hospital for a very wide area. During the same period, among 9565 admissions to the surgical wards there were 200 of cancer of the breast, a ratio of 1 to 37*8. And among 5370 gynaecological admissions (for a slightly shorter period) there were 221 of cancer of the uterus, a ratio of 1 to 24-3.

Etiology and Incidence. — Sex. There are varying statements as to the relative frequency in the two sexes. Thus Brinton considered the disease to be twice as frequent in males. Welch1 among 2214 cases found 1233 males and 981 females. In a larger series of 5426 there were 2843 males and 2583 females. Reiche 2 reports the statistics from Hamburg for the years 1872 to 1898. During this period there were 4237 deaths from cancer of the stomach, of which 2387 were in males and 1850 in females, a proportion of 1*8 to 1. Of 772 cases of cancer of the oesophagus, 609 were in males and 163 in females, a proportion of 5'2 to 1. Of our series 126 were in males and 24 in females, a proportion of 5 to 1.

Age. — The following table shows the number of cases occurring in each decade in our series, and for comparison the percentages of Welch's table of 2038 cases and Reiche's Hamburg statistics.

1 Loe. cit.

Age.

Number of Cases.

Percentage.

Percentage of Welch's Table.

Percentage of Reiche's Table.

10-20

•1

•04

20-30

2-7

•87

30-40

11-3

13-3

5-85

40-50

14-15

50-60

32-6

30-4

25-94

60-70

32-16

70-80

2-6

6-8

18-13

Over 80

1-15

2-76

It is seen that the Hamburg statistics give the maximum figures in the seventh decade, and that in the earlier periods the percentages are less than in "Welch's series. This was the case both in males and females. The large percentage of cases between the years 70 and 80 is striking. Of our cases below the age of 30, there was one each of the ages of 22, 24, 25, and 26, with two of the age of 28. The oldest cases in our series were two of 78. It will be seen that 57*9 per cent, were between the ages of 40 and 60.

In the census report for 1890, the death-rate from cancer of the stomach per 100,000 living in the registration areas is : —

All Ages.

Over 65 Years.

10-24

3-22

34-45

79-96

The large figure for the population over 65 years of age is of course due to the smaller number of persons living.

Race. — The relative numbers in the white and coloured patients were, white 131 and coloured, 9 — a ratio of 6*9 to 1. The relative proportion in the hospital is about as 6 white to 1 coloured.

The question as to the occurrence of cancer among the coloured race is of much interest, especially in connection with the statement as to the infrequency of carcinoma among the negroes in Africa. In our coloured patients it is impossible to speak positively as to the amount of admixture of white blood.

The nationality of the patients is given in the table below, in which the relative percentage of each nationality among the total admissions to the hospital is also given : —

Birthplace.

Cases in

Percentage in

Percentage of

Series.

Series.

all Patients.

United States .

57-3

Germany .

26-6

11-3

Ireland

5-3

2-8

England .

1-4

Various .

6-6

5-4

Thus while the native-born supply 78' 3 per cent, of all patients they supply only 57*3 per cent, of the cases of cancer of the stomach, and the 21-7 per cent, of foreign-born patients supply 42*7 per cent, of the cases. This is probably due to the fact that a very large percentage of the foreign-born are adults, and so supply a relatively large number of individuals in the "cancer-age." The census returns in regard to colour and nationality are of interest. Thus in 1880 in the registration area the figures of deaths from cancer generally per 100,000 living were : —

White Males.

Coloured Males.

White Females.

Coloured Females.

20-54

5-85

35-44

19-32

This area contained a population of 29,000,000, of which 6,000,000 were coloured. The death-rate from all cancers per 100,000 living was for the whites 27*96, and for the coloured 12-67. As to nationality the rate per 100,000 living among the native-born was 20*08, and among the foreign born 53-30. Mullins1 has noted the large ratio of deaths from cancer among the foreign-born population in certain parts of Australia. Those born in Britain, while only composing 23'5 per cent, of the population, famish over 71 per cent, of the cases of cancer. The German born, who are only 0-85 per cent, of the population, furnish 3-4 per cent, of the deaths from cancer.

Regarding the deaths from cancer of the stomach in the registration areas, in 1890, the rate per 100,000 living for the white native-born is 5-84 for all ages, and 22*26 for the white

1 Log. cit.

CANCEK OF THE STOMACH foreign-born. A lower death-rate is found also in the children of mothers born in the United States, the highest being in the children of mothers born in Germany. This is most marked in the cases over sixty-five years of age, being 47-72 in children of native-born mothers, and 123-62 in the children of mothers born in Germany.

These figures indicate a lower death-rate from cancer in the coloured race as compared with the white, and in the white native-born as compared with the foreign -born.

The age, sex, and colour incidence are given in the table : —

Age.

White Males.

White Females.

Coloured Males.

Coloured Females.

Total.

27 4

0 5 5 6 6 0

2 3 4 5 3 0

0 0 0 2 0 0

6 17

Total

Conjugal Condition. — There were 137 married or widowed, and 13 single. In the census returns for 1890 the figures per 100,000 living of deaths due to cancer in regard to the conjugal relation were : —

Condition.

Males.

Females.

Single .

Married Widowed

17 100 265

Family History. — In connection with this series, for the sake of comparison, the previous history of 150 cases of adults with the same relations of sex and colour was noted. These were taken at random among the medical cases, and are not intended to express any constant ratio. They are of interest, however, to compare with the cases of the series.

Statistics as to the relative frequency of a family history of cancer vary widely, from 1 in 3 to 1 in 28. Welch states that probably in about 14 per cent, it can be determined that other members of the family are or have been affected with the disease.

This estimate is based on an analysis of 1744 cases, in 243 of which a family history was determined. There were 11 patients, or 7*3 per cent., among our series giving a family history of cancer. Of these the special organs affected were : 2 — one parent, cancer of stomach ; 2 — one brother or sister, cancer of stomach ; 2 — mother, cancer of uterus ; 2 — father, cancer of face ; 3 — sister, cancer of breast.

Among the comparative series there were 6 cases with a family history of carcinoma, one of these being of the stomach. Our figures do not support the statement that the disease developes at an earlier age when there is a family history of cancer. The following table shows the ages of the cases with a family history of cancer : —

Age.

Number of Cases.

Cases with a Family History of Cancer.

0 2 6 3 0

So that among the 23 cases which occurred below the age of 40, there was not one with a family history of cancer. The majority were among the cases in the decade with the greatest number of cases of cancer.

The influence of heredity is one about which widely different opinions are expressed. Certain exceptional cases of many members of the same family having had cancer — e.g., the Bonajjartes — have largely influenced opinion. Many of the French writers especially hold the view of an inheritance of cancer or of a tendency thereto, and in this sense use the term " la diathese neoplastique." It is natural that in certain areas in which cancer appears to be common, as for example in Normandy, the occurrence of several cases in the same family should seem to support the probability of heredity. However, the Registrar-General's Report for 1889 points out (page xiv.) that as 1 out of 21 males and 1 out of 12 females who reach the age of 35 years die of cancer, it follows that on an average in 1 out of 3 cases either a parent or a grandparent will have died of this disease — if they died aged 35 years or over. Statistics from private practice are likely to be of greater value than those gathered from hospital patients, who often know little about any but their more immediate family.

Tuberculosis. — There were 38 patients who gave a family history of tuberculosis, or 25*3 per cent. It is a coincidence that in the comparative series there were exactly the same number with a tuberculous history. The ages of these cases are given in the following table : —

Age.

Number of Cases.

Cases with a tuberculous history.

6 17 38 49 36

As in the patients with a cancerous family history, these do not apparently occur at an earlier age on account of the tuberculous family history. It may be stated that in 41 autopsies in this series there were tuberculous lesions in 11, of which 6 were old and 5 recent.

Lues. — Only 9 patients gave a history of syphilis.

Previous Stomach Trouble. — Until the appearance of Ewald's work on the stomach we had always held the view, based on nothing but vague impressions, that the subjects of chronic dyspepsia were much more prone than others to malignant disease of the stomach. In the present series this question has been inquired into with special care. It is of interest in connection with, first, the possibility of the development of a cancer on the site of a simple ulcer, and secondly, the supposed causal association between an acute or chronic gastritis and subsequent cancer.

In only four of our series was there a history which pointed to gastric ulcer, and in none of these could such a condition be positively diagnosed. In none of the autopsies was this condition found.

Thirty-three patients gave a history of stomach trouble before their present illness. Of these, 17 had isolated attacks — often many years before, and in some only single — which generally seemed to have been an acnte gastritis ; 11 had chronic stomach trouble for some years, and in some it had existed during most of their life ; 5 had stomach trouble, more or less constant, for one or two years before the present symptoms became pronounced.

It is interesting to note the incidence of stomach troubles in the series of hospital cases which we have taken for comparison. Twenty-eight of them gave a history of old gastric trouble ; in 23 it had lasted for many years, while 5 of them had had also occasional acute attacks. This is a percentage not very much smaller than in the cancer series, so that we may conclude, so far as the figures show anything, that the victims of chronic dyspepsia and the various forms of gastritis are not more prone to malignant disease than other individuals.

Alcohol. — About one-half of the patients— 77— had used alcohol regularly. Of these 8 were described as heavy, 65 varied from moderate to heavy, and 4 were light users. In the comparative series there were 65 who had been regular users of alcohol, of these 16 were heavy drinkers, 38 varied from moderate to heavy, and 9 were light users.

Miscellaneous Conditions. — Mental influences, as worry, strain, exposure and hardship, have been variously estimated as causal factors. Gallet and Deschamps,1 in an inquiry into 722 cases of cancer, found that among 67 instances of cancer of the stomach, there were 44 giving a history of depressing mental conditions. Only isolated cases in our series suggested such a connection, and they are not numerous enough to be taken into account.

Local Irritation.- This offers a tempting field for investigation, but there are no specially conclusive facts to support such a view. Possible irritation from food or drink is urged by some writers. Thus in certain parts of Normandy, where cancer of the stomach is unusually prevalent, Brunon 2 and Rebulet 3 attribute this largely to the consumption of a special kind of cider made with sea water. As many of the cases occurred in the same family or in the same houses, these writers have also attributed an influence to heredity and contagion. The drinking of very

hot strong tea has been advanced as a cause, and also the eating of large amounts of meat. There were not a sufficient number of cases in this series to show any causal agency in the previous food or drink.

Trauma. — Some writers, as Boas, lay a good deal of stress on this as an etiological factor. That an accident or injury should be sometimes regarded as the starting-point of the disease is not remarkable. In certain cases the malignant disease is latent, and the trauma is the factor which brings out the symptoms. When the onset is sudden this is likely to be mentioned as a cause. In several of the cases in this series an indulgence in very cold beer was referred to by the patients as a possible cause. In only a single instance in the 150 cases was there apparently a direct connection between the trauma and the onset of the disease. The following case illustrates the abrupt onset of stomach symptoms after an injury : —