- Email: [email protected]
A Better Prognosis for Obese Men with Prostate Cancer
00225347/96/15514220$03.~ 'hm JOURNAL OF'U M ~ Y Copyright 0 1996 by -CAN U M ~ I C NASSOCIATION, . hc.
Vol. 155,220-225, January 1% Printed in U.S.A.
A BETTER PROGNOSIS FOR OBESE MEN WITH PROSTATE CANCER HARRY W. DANIELL* From the Department of Family h t i c e , University of California Medical School at Davis, Redding, California
ABSTRACT
Purpose: The possibility was investigated whether obesity is associated with a favorable prognosis in men with nonstage A prostatic cancer independent of other risk factors. Materials and Methods: A total of 235 men with nonstage A prostatic cancer diagnosed between 1983 and 1990 at a community hospital was evaluated by tumor grade and stage, degree of obesity, smoking habits, age, year of diagnosis and survival. Results: In comparison to tumors of less obese men those in men who were at least 10% overweight were more often stage B or C than D (odds ratio 2.3,95% confidence interval 1.4to 4.9). Smoking abstinence, patient age older than 70 years and low Gleason scores were also independently associated with less advanced tumors. The lower frequency of advanced stage tumors among those of obese men was most prominent for stage D2 disease in nonsmokers, progressively decreasing from 24% of 29 nonsmokers who had been underweight for many years before diagnosis to 8%of 24 who had been greatly overweight for many years (p <0.05). Tumor specific mortality was also lower in obese men and nonsmokers independent of tumor stage and grade with 5-year tumor specific mortality rates of 10%in obese nonsmokers, 27% in nonobese nonsmokers, 36% in obese smokers and 43% in nonobese smokers. These differences reflected a combination of obesity related differences in stage distribution and a more favorable prognosis for obese men among patients with tumors of similar stages. Conclusions: These observations suggest inhibition of prostate cancer growth and metastasis by increased endogenous estrogen, decreased endogenous testosterone or other systemic changes characteristic of obesity. Kcr WORDS:prostate, prostatic neoplasms, obesity, smoking, neoplasm metastasis
Invasion and metastasis by prostate cancer involve interplay between tumor aggression and host antitumor defenses. Contributions by these tumors to disease progression have commonly been evaluated by relating their characteristics to cancer stage at diagnosis, or rates of recurrence or tumor specific mortality. Interference with tumor progression by antitumor therapy has been evaluated by analysis of recurrence and mortality rates, or quantitation of tumor size or tumor generated biological markers. Associations between disease progression and host factors have rarely been reported although great differences in mortality from prostate cancer are recognized between populations in which the incidence of these tumors is apparently similar. Prostate cancer of more advanced stage and worse overall prognosis have been associated with younger age1 and African ancestry,2.3 relationships that are usually thought to have resulted from delayed diagnosis. In 1 large series prostate cancer prognosis was unrelated to patient age after controlling for tumor grade.4 Relationships between obesity and prostate cancer have not been well defined. Studies in different populations demonstrated that the frequency of these tumors in obese compared to nonobese men varies from a 4fold increases to less6v7 or no differen~e.~.S None of these studies analyzed cases by tumor stage. Both prospective series relating obesity to mortality from prostate cancer indicated a 20 to 35% increase in mortality in obese 11 These observations are compatible with a higher frequency of but more favorable prognosis for prostate cancer in overweight men. The growth and metastasis of most prostatic cancer are inhibited by estrogen supplementation or androgen deprivation. Since obese men characteristically have increased endogenous estrogen and decreased endogenous testosterone Accepted for publication July 21, 1995. * Requests for reprints: 2626 Edith Ave., Suite A, Redding, California 96001.
levels,12-14 the possibility of tumor inhibition induced by these altered sex hormone levels was evaluated by designing a study that included quantitation of obesity. This study examined the associations between several host factors, and tumor stage and tumor specific mortality in white men with nonstage A prostatic cancer.16 Obesity was independently associated with less advanced disease at diagnosis, as was smoking abstinence, older age and low tumor grade. Details of the associations between smoking, and tumor stage and prognosis have been previously described.16 "he present report analyzes data from this cohort, which relate obesity to tumor stage and tumor associated death. Fewer advanced tumors among those of obese men indicate less aggressive early disease progression unless there is an earlier diagnosis or tumor induced weight loss. A more favorable prognosis in obese men after controlling for tumor stage and grade, age and smoking habits would indicate less aggressive late disease progression. MATERIALS AND METHODS
The hospital records were reviewed of all 240 men with an initial diagnosis of nonstage A prostate cancer between July 1983 and December 1990 at Mercy Medical Center, 1 of 2 community hospitals serving a semirural, largely white population at the northern end of the central valley of California. Recorded data included year of diagnosis, patient age, height, weight, smoking habits, presenting signs and symptoms, prostate surgery performed, stage and Gleason grade at diagnosis. Survival data were obtained from personal physicians, hospital cancer registry and death certificates. Recent weight changes in stage D cases were calculated by comparing recorded hospital or office weights at the time of diagnosis with documented hospital or office weights recorded 1 to 4 years before cancer diagnosis. In 5 cases earlier weights were not
220
OBESITY AND PROSTATE CANCER PROGNOSIS
22 1
TABLE1. Characteristics of235 nonstage A prostate cancer patients analyzed by age at tumor diagnosis F’t. Aee at Diaenosis I
Younger Than 60
No. pts. Pt. characteristics (Yo): Smoking 1109 Recommended wt. or more 120% Recommended wt. or more 41, or More weight loss Pelvic lymphadenectomy Tumor characteristics: % Stage D1 ‘7r Stage D2 Av. grade* 1 5-Yr. tumor specific mortality * Not available for 3 tumors.
6044
-
0-99 100-109 110-119 120-129 130-139 140orGreater Totals
44 42 53 56 36
(21) (27) (31) (19) (5)
53
66
46
19
235
65 53 29 12 59
29 50 35 12 68
26 64 30 4 70
26 64 33 5 67
13 26 13 9 41
5 47 16 16 0
25 52 27 8 57
24 29 5.18 24
29 18 5.34 22
19 21 5.80 32
15 15 5.44 18
7 22 5.59 22
0 21 5.71 31
16 20 5.54 23
Total 70
(NO. pU.1
D1
D2
6 (3) 17(11) 24(14) 21 (7) 21 (3)
19 (9) 20(13) 7 (4) 18 ( 6 ) 14 (2)
44(15) 22(l4) 17 (9) 6 (2) 29 (4)
100 (48) 100 (65) 100 (58) 100 (34) 100 (14)
17 (41)
16 (37)
20 (46)
100 (235)
older), recommended weight (less than 109% versus 110% and greater) and smoking status. Probabilities of empirical results were determined by the chi-square test under the null hypothesis. Death certificates of 100 of the 104 men known to have died were reviewed. A total of 57 certificates listed prostate cancer as the primary cause of death and 7 included prostate cancer as a contributing cause. These 64 deaths were tabulated as prostate cancer deaths. The 4 men with an unknown cause o f death were tabulated as alive on the date preceding death. RESULTS
Table 1 shows tumor and patient characteristics. Older men were less often smokers or obese, and less likely to have undergone pelvic lymphadenectomy and, therefore, to have had stage D1 tumors. Factors unrelated to patient age included the percent of stage D2 tumors, tumor grade and 5-year tumor specific mortality. Table 2 lists the percent of men in each obesity group who presented with disease of each stage. Stage D2 tumors were less common among those of men in most weight categories of greater than 110% of recommended weight. Previously reported multivariate analysis demonstrated independent associations between stage D disease and lack of obesity (odds ratio 2.3,95%confidence interval 1.4 CO 4.9, p = 0.003),Gleason grade 7 to 10 (p = 0.001),patient age younger than 70 years (p = 0.008) and smoking (p = 0.015).’5 The association between stage D disease and younger age was largely but not completely due to the much higher frequency of pelvic lymphadenectomy in younger men (table 1). Gleason scores were unrelated to smoking habits or obesity (table 3). Table 4 compares weights recorded 1 to 4 years before diagnosis with those recorded at diagnosis, and indicates changes in weight categories during the interval between these observations. Ten patients (24%)with stage D2 disease lost more than 6% of body weight but only 4 had changed from above to below 110% recommended weight. Of the 6 patients whose category changed from obese to nonobese 3 had lost 7 or fewer pounds. Weight loss before diagnosis was TABLE3. Average Gleason scores for men with nonstage A prostate cancer Av. Score (No. tumors)* % Recommended Wt.
Smokers
0-99 100-109 110-119 12LL129 130 or Greater
Totals 47 (111)
Overall Values
.
34
% Tumor Stage (No. pts.)
C
Than80 Older
75-79
17
TABLE 2. Tumor stage and degree of obesity in men with nonstage A prostatic cancer B
70-74
~~
available and in 3 weights at diagnosis were confounded by the presence of edema. In these 3 patients dry weights measured l to 4 years before diagnosis were used in calculations. The smoking history of 184 men was obtained from the oftice or hospital records of urologists who required patients to complete forms detailing personal habits as part of the preoperative evaluation, and for most others from intake interviews by the nursing service, history documented by anesthesiologists or other consultants, or office records of attending urologists or personal physicians. Failing adequate data from these sources the history of tobacco use was obtained from 11 men or their families by the office staffs of their physicians. Height or weight of 5 men was unavailable, leaving 235 for analysis. Included were 3 patients with stage D2 disease for whom tumor grade was unavailable because the small amount of malignant tissue identified did not allow for grade determination (l),or biopsies were not done after presentation with multiple bony metastases and markedly elevated prostate specific antigen determinations (2). Men who had smoked a t least 10 cigarettes (1/2 pack) for a t least 5 years, including a portion of the 3 years before surgery, or who were currently documented smokers of cigarettes, a pipe or cigars were tabulated as smokers. More detailed analysis of smoking habit and its associations with tumor stage and tumor specific mortality has been previously reported.15 Obesity was quantitated by dividing recorded weight by the average weight for recorded height and medium habitus recommended in the 1983 Metropolitan Life Insurance height and weight tables, resulting in the percent recommended weight. Histological examination of surgical specimens was perfarmed by a group of pathologists who freely collaborated on difficult or unusual cases. Tumor grade was assigned by these pathologists after examining biopsy and staging tissues. If biopsy and prostatectomy tissues differed in tumor grade, the higher grade was tabulated. Stepwise logistic regression analysis was performed for associations between stage D tumors and Gleason score (2 to 6 versus 7 to lo), host-patient age (younger than 70 years versus 70 years and
1 Recommended Wt.
65-69
* Not available for 3 tumors.
6.1 (17) 5.5 (17) 4.8 (12) 5.0 (7) -6 5.5 (58)
Nonsmokers 5.1 (30) 5.4 (48) 5.5 (451 6.0 (26) 5.7 (25) 5.5 (174)
OBESITY AND PROSTATE CANCER PROGNOSIS
222
TAEW 4. Weight changes preceding diagnosis of stnge D prostatic cancer ~~
~
Sbge
D1
Stage D2
No. No. No. No. Smokers Nonsmokers Smokers Nonsmokers
0
25 1 0
17 1 2
10
24
14
1 27
0 0
0 2
3
3
0
8 1
7 0
1
18 2 1
0 15
4
3 6
0
0
0
2
2
0
0
4
0
0
1
0
1
0
1
2
12
No. pta. No data
2
Disquafified (edema)
Documented aerial wta. Change in % rearmmended
29 1
E
wt.:
Greater than 6% gain 44% Gain Less than 4% gain or loss
44% Loss Greater than 6%loss change in obesity categoly
049%
100109% 110-119% 120-12!4%
130%+
-
FIG. 1. Percent of nonstage A prostate cancers presentingat stage D2 in nonsmokers ouped by degree of patient obesity before tumor associated weight %ss. Values in parentheses indicate number of men.
110%
Greater than to leas than
'
PERCENT RECOMMENDEDWEIGHT 1 4 YEARS BEFORE DIAGNOSIS
before diagnosis:
Less than to greater than
0%
110%
Lesathantogreatertllm 120%
Greater tban to less than
unrelated to obesity in smokers or nonsmokers (data not shown). Of men with stage D2 disease those older than 75 years had more often lost weight recently than younger men (50% versus 19%, p <0.03). Table 5 lists the distribution of obesity categories in smokers and nonsmokers with stage D2 disease before and after tumor associated weight changes. No relationship is apparent in smokers between obesity and the presence of D2 disease. In nonsmokers a pattern of less frequent stage D2 disease in more obese men was suggested long before tumors were diagnosed (p
% Rscommended
wt.
and tumor stage. No relationship was suggested by the 5-year tumor specific mortality between younger and older men with similar smoking habits and obesity, an analysis handicapped by the small numbers in some age groups. Better &year tumor specific survival rates of obese men persisted in smokers and nonsmokers after correction for minor differences in age distribution. Tumor specific mortality was lower in obese than nonobese nonsmokers (p <0.01) and in obese nonsmokers than all others (10% versus 30%, p <0.001). Tumor specific survival rates for obese and nonobese smokers and nonsmokers with stage D disease, and all patients with nonstage A tumors are shown in figure 2. The longer survival of nonobese nonsmokers reflects a combination of lower mortality in those with tumors of identical stages (table 8) and a larger percent with less advanced tumors. Patients had received a wide variety of therapies, most of which were conventional, and administered with varying intensity and consistency. Some patients were lost to followup until death. No differences in antitumor therapy were noted in men of similar ages but with different smoking habits or degrees of obesity. DISCUSSION
These observations are compatible with the inhibition of prostate cancer growth, invasion and metastasis by factors associated with obesity, which are most prominent in nonsmokers and independent of other risk factors. Since many prostate cancers are uniquely sensitive to sex hormone manipulation, and to my knowledge a similar relationship has not been demonstrated in patients with other types of malignant tumors, the estrogen and testosterone changes characteristic of obesity seem likely to be responsible for these associations. The possibility that earlier diagnosis of prostate cancer in obese men contributes to the lower percent of advanced tumors at diagnosis seems remote. Rectal examinations are more difficult to perform in men with obese buttocks, which
TABLE5. Smoking habits and a'egree of obesity in men with stage 0 2 prostate cancer 1-4Yrs.Before Diagnosis. At Diagnosis 96 Smokers (No.)
96 Nonsmokers (No.)
% Smokers (No.)
0-99
35 (17)
100-109 110-119 120-129
13 (16) 8 (12)
24 (29) 20 (46) 14 (44) 10 (30) 8 (24)
39 (18) 18 (17) 15 (13)
38 18)
Greater than 130 50 (4) Not available for 3 smokers and 2 nonsmokers.
% Nonsmokers (No.)
29 (7) 60 ( 5 )
12 (25)
OBESITY AND PROSTATE CANCER PROGNOSIS T-LE p/r
223
6 . Smoking habits and degree of obesity in men with stage D l prostate cancer who underwent pelvic lymphadenectomy
Recommended Wt.
1-4 Yrs. Before Diagnosis* % Smokers (No.)
0-99 33 (6) 100-109 40 (10) 110-119 11 (9) 120-129 50 (4) Greater than 130 50 (2) * Not available for 2 smokers and 1 nonsmoker.
B Nonsmokers (No.)
36 (11) 24 (29) 15 (27) 25 (20) 33 (12)
TABLE7 . Men with nonstage A prostatic cancer diagnosed earlier (1983 to 1986) and later (1987 to 1990)
At Diagnosis 8 Smokers (No.)
9% Nonsmokers (No.)
50 (8) 50 (12) 0 (8) 33 (3) 50 (2)
42 (12) 24 (29) 15 (27) 25 (20) 33 (12)
indicate differences in the frequency or intensity of smoking, or undefined dietary, genetic, environmental or therapeutic differences, some of which may reflect the decades between NO. RS. (B stage D c a ) the previous and present observations, obesity related differ1983-1986 1987-1990 ences in the incidence of nonstage A prostate cancers andlor Obese nonsmokers 37 (24) 61 (23) statistical aberration. Nonobese nonsmokers 33 (48) 44 (34) Relationships between obesity and the incidence of prosObese smokers 16 (63) 8 (25) tate cancer are incompletely defined. Talamini et al reported Nonobese smokers 21 (62) 15 (47) on a population of men in northern Italy in which the inciTotals 107 (44) 128 (29.7) dence of prostate cancer in obese men was greater than 4 times that of men of the lowest quartile in percent body weight.5 A similar obesity related pattern of incidence in the frequently do not allow a thorough search for prostate nod- present population would make the data from this cohort ules. This pattern would be expected to delay the diagnosis of easily compatible with those reported by Snowden et all1 and palpable asymptomatic malignancies in some men. Obese Lew and Garfinkel.10 However, others noted a 20 to 35% men undergo less surgery for benign prostatic hypertro- increase in prostate cancer incidence6.7 or no increase*.g in phy16.'7 and have larger adenomas a t operation,la suggesting obese men. None of these incidence studies reported tumors later onset of obstructive symptoms in those with benign by stage, thereby failing to distinguish between potentially enlargement and longer delay in the diagnosis of some tu- lethal tumors and those of little clinical concern. The degree mors identified during the evaluation of symptomatic benign to which obesity related differences in prostate cancer incidisease. The possibility of inherently fewer aggressive malig- dence between these populations are due to differences in nancies among those of obese men as an explanation for the techniques of tumor identification or environmental, geolower percent of advanced tumors or lower tumor specific graphic and other factors largely remains to be investigated. mortality is not supported by the similarity of tumor grades No incidence studies have controlled for smoking habits, in the obese and nonobese men in this study. although our recent studies suggest a higher incidence of Tumor associated weight loss apparently contributed only stage A diseasezzand lower incidence of nonstage A disease15 slightly to the aforementioned associations although the fre- in smokers. Different risk factors for aggressive and nonagquency and degree of weight loss were similar to those in gressive prostate cancers are also suggested by the dietary larger studies. Of the patients with stage D2 disease in the studies reported by West et al.23 present series 28% had lost 4% or more recommended body The more favorable prognosis for obese than nonobese nonweight compared to 27% of 1,020 men with stage D2 dis- smokers with tumors of identical stages in the present series easels and 28% of 78 men with similarly advanced tumors in suggests that a t least some of the influence on disease proother series.2" In the only other study associating weight or gression associated with obesity may continue to function obesity with death rates in men with prostate cancer Byar even in the presence of conventional antitumor therapy. If and Corle reported fewer deaths from any cause in 80%of the this pattern is confirmed by future studies, it suggests that men with advanced prostate cancer who weighed more than conventional therapy may not offer the optimal therapeutic 59 kg. (130 pounds) than the 20% who weighed less. HOWever, no association was noted between body weight and combination of sex hormone supplementation and inhibition death attributable to prostate cancer, and they did not in- for many patients with advanced prostatic cancer, and that clude analysis by tumor stage or tumor associated weight optimal antitumor therapy may differ in men of different obesity status or smoking habits. loss.21 In the present series the deaths of almost all men with Both prospective studies that assessed mortality rates stage D disease who died during followup were due to prosfrom prostate cancer in men with different degrees of obesity demonstrated increased death rates from these tumors in tate cancer. These data suggest that obese nonsmokers with obese men. Lew and Garfinkel reported cause-specific mor- stage D disease may live longer if they maintain excess tality rates for many tumors in 336,442 men older than 30 weight despite increased death rates from cardiovascular Years enrolled for 12 years beginning in 1960.1° Men with a and other nonmalignant diseases associated with obesity. In recommended weight of 110 to 119%, 120 to 129%, 130 to this cohort deaths from all causes were less frequent in obese 139% and greater than 140% had age adjusted prostate can- than nonobese men with prostate cancer independent of tucer specific death rates of 0.9, 1.37, 1.33 and 1.29%, respec- mor stage. Appropriate additional studies will be of interest. Although estrogen and testosterone levels were not availtively, compared to 1% for those of 90 to 109%recommended able in the present study, data from other reports indicate weight. Smoking habits were not analyzed. Snowdon et a1 reported higher mortality from prostate cancer in 6,763 men that estrone production in men who are 30% more obese than weighing more than 130% followed for 20 years beginning in the recommended weight may approximate 120% of that in 1960.11 However, this obese group only included 13 deaths men of ideal weight,13,l4 an increase in estrogen influence from prostate cancer, the analysis did not include smoking equal to approximately that of0.12 mg. diethylstilbestrol per habits and men with a recommended weight of 110 to 129% day, and that testosterone production in these obese men had a death rate from prostate cancer that was almost iden- may have decreased to between 85% and 90% of that of tical to that of less obese men. Differences between t h e men of similar age,12Bi3changes of a lower magnitude than Present results and those of the 2 preVious Studies may those induced by conventional hormonal therapy. These o h -
OBESITY AND PROSTATE CANCER PROGNOSIS
224
TABLE8. 5-Year cumulative mortality in men with nonstage A prostate cancer according to tumor stage, smoking habits and obesity before tumor associated weight lossx All Cause Mortality
Tumor SpecSc Mortality Tumor
% Smokers
% Nonsmokers
% Smokers
stage
(No. men)
(No. men)
(No. men)
Obese
Nonobese 90 (10)
02
26 (8) 16 (15) 42 (33) *Weight at diagnosis in 8 men.
BandC E D
A
Obese
Nonobese
Obese
Nonobese
Obese
65 119) 27 (12) 8 (46) 25 (77)
53 110) 8 (13) 5 (75) 10 (98)
90 (10) 26 ( 8 ) 41 (15) 52 (33)
86 (7) 25 (4) 14 (16) 34 (27)
68 (19) 27 (12) 24 (46) 35 (77)
61 (10) 27 (13) 15 (75) 21 (98)
B
PROSTATE CANCERS
1
2
3
4
5
6
PROSTATE CANCERS
7
1
2
YEARS
4
5
6
7
7
MEN REMAINING
-
10
7
7
3
3
3
2
27
25
22
21
16
12
9
18
15
12
11
10
6
6
3
*----*33
29
25
24
20
13
11
7
25
23
22
19
14
12
5
3
98
97
92
84
66
49
25
15
29
24
17
17
12
10
9
6
-----a 77
65
57
54
44
34
23
17
-11
O----O
3 YEARS
MEN REMAINING *-----O
(No. men)
Nonobese
86 (71 25 (4) 14 (16) 34 (27)
D1
% Nonsmokers
FIG.2. A, tumor specificsurvival of obese and nonobese smokers, and nonsmokers with stage D prostate cancer. Obesity status preceded tumor associated weight loss or gain. B, tumor speciSc survival of obese and nonobese smokers, and nonsmokers with nonstage A prostate cancer. sity related differences in hormone production were established in men younger than 60 years and may be greater or less than differences in those in older age groups, which include most men with prostate malignancies. Others have suggested modification of the behavior of breast cancers, another group of sex hormone sensitive malignancies, by the obesity associated changes in these hormones.=.% In women with breast cancer obesity is associated with more advanced disease at diagnosis,26 larger axillary metastasis27 and a less favorable prognosis after controlling for tumor stage and other fa~tors.24~25 In the present study the greater improvement in prognosis associated with obesity in nonsmokers suggests that the mechanism by which this improvement is effected may be neutralized in smokers. This pattern may reflect the increased 2-hydroxylation of estradiol demonstrated in smokers by Michnovicz et al,m which has been proposed as the mechanism responsible for the antiestrogen influence promoting premature menopause, osteoporosis and other estrogen deficiency diseases in women who use tobacco. The prominent stage migration after 1986 in the present series and those of others has been thought to occur on the basis of earlier tumor diagnosis in later years resulting from more widespread use of prostate specific antigen determinations and a more aggressive pursuit of early prostate cancer diagnosis by physicians and patients. However, in the present study the more frequent obesity and less frequent smokingby patients in whom the disease was diagnosed after
1986 were responsible for approximately an eighth of the stage migration in this cohort. If smoking habits and degree of obesity had remained the same after 1986 as before that date and the percent of advanced tumors in each smokingobesity category had retained the improvement noted during the later interval, the percent of those with stage D disease diagnosed after 1986 would have been 31.5% rather that 29.7%, compared to 44.9% before 1986. Similarly, stage migration contributed little to the association between obesity and less advanced tumor stage. Of these men 28.7% who were and 45.1% who were not obese presented with stage D disease. If the percent of men diagnosed after 1986 had been identical in obesity and smoking status to that before 1986 and the lower percent of advanced tumors had been the same, 29.2% of obese and 45.7%of nonobese men would have presented with stage D disease. CONCLUSIONS
Many studies examining mortality from prostate cancer have shown that a 5-year followup is insufficient to demonstrate differences that are clearly present during longer intervals due to the slow progression of the majority of prostate malignancies. Although much of the improved 5-year tumor specific mortality rate associated with obesity in the present cohort was due to tumor related differences in stage distnbution (table 7), the prognosis of obese nonsmokers with stage D disease was clearly more favorable than that of
OBESITY AND PROSTATE CANCER PROGNOSIS
nonobese nonsmokers with similar tumors (fig. 2, A). Longer followup should clarify the relative importance of obesity related differences in stage distribution and tumor behavior after diagnosis to the improved survival of obese men with prostate cancer. The present technique for establishing smoking habits, largely obtained from chart review, includes considerable potential error, which has been discussed previously.15 The similar technique for identifying the height and weight of each patient in the present study also includes the potential for error but data similar to those tabulated had usually been documented by different recorders in multiple office and hospital chart locations, and often during different periods of hospitalization. These weaknesses in study design seem unlikely to have modified the described associations. Prospective studies using more conventional techniques for quantitation of tobacco use and comparison of the prognosis of men who stop smoking or gain weight after diagnosis to that of men who do not will be of great interest. Dr. T. Joseph Sheehan performed the multivariate analysis, Billie White prepared the graphics and Phyllis Watkins and Valerie McRae assisted in patient followup. REFERENCES
1. Hamson, G. S. M.: The prognosis of prostatic cancer in the younger man. Brit. J. Urol., 5 5 315, 1983. 2. Brawn, P. N., Johnson, E. H., Kuhl, D. L., Riggs, M. W., Speights, V. O., Johnson, C. F., 111, Pandya, P. P., Lind, M. L., and Bell, N. F.: Stage at presentation and survival of white and black patients with prostate carcinoma. Cancer, 71: 2569, 1993. 3. Polednak, A. P. and Flannery, J . T.: Black versus white racial differences in clinical stage a t diagnosis and treatment of prostatic cancer in Connecticut. Cancer, 7 0 2152, 1992. Damber, J.-E., Jonsson, H. and Lenner, P.: Patient 4. Gronberg, H., age as a prognostic factor in prostate cancer. J. Urol., 152:892, 1994. 5. Talamini, R., La Vecchia, C., Decarli, A., Negri, E. and Franceschi, S.: Nutrition, social factors and prostatic cancer in a Northern Italian population. Brit. J . Cancer, 53: 817,1986. 6. Nomura, A., Heilbrun, L. K. and Stemmermann, G. N.: Body mass index as a predictor of cancer in men. J. Natl. Cancer Inst., 74:319, 1985. 7. Mills, P. K., Beeson, W. L., Phillips, R. L. and Fraser, G. E.: Cohort study of diet, lifestyle, and prostate cancer in Adventist men. Cancer, 64:598, 1989. 8. Kolonel, L. N., Yoshizawa, C. N. and Hankin, J . H.: Diet and prostatic cancer: a case-control study in Hawaii. h e r . J. Epidemiol., 127: 999, 1988. 9. Graham, S.,Haughey, B., Marshall, J., Priore, R., Byers, T., Rzepka, T,, Mettlin, C. and Pontes, J. E.: Diet in the epidemiology of carcinoma of the prostate gland. J . Natl. Cancer Inst., 7 0 687,1983. 10. h w , E . A. and Garfinkel, L.: Variations in mortality by weight among 750,000 men and women. J. Chron. Dis., 3 2 563,1979. 11. Snowdon, D. A,, Phillips, R. L. and Choi, W.: Diet, obesity and risk of fatal prostate cancer. Amer. J . Epidemiol., 120 244, 1984. 12. Zumoff, B., Strain, G. W., Miller, L. K., Rosner, W., Senie, R..
225
Seres, D. S. and Rosenfeld, R. S.: Plasma free and non-sexhormone-binding-globulin-bound testosterone are decreased in obese men in proportion to their degree of obesity. J. Clin. Endocriol. Metab., 71: 929,1990. 13. Simon, D., Preziosi, P., Barrett-Connor, E., Roger, M., Saint-Paul, M., Nahoul, K. and Papoz, L.: The influence of a g h g on plasma sex hormones in men: the Telecom Study. Amer. J. Epidemiol., 136 783, 1992. 14. Kley, H. K,Deselaers, T., Peerenboom, H. and Kruskemper, H. L.: Enhanced conversion of androstenedione to estrogens in obese males. J . Clin. Endocrinol. Metab., 51: 1128, 1980. 15. Daniell, H. W.: A worse prognosis for smokers with prostate cancer. J . Urol., 154:153, 1995. 16. Sidney, S.,Quesenberry, C., Jr., Sadler, M. C., Lydick, E. G., Guess, H. A. and Cattolica, E. V.: Risk factors for surgically treated benign prostatic hyperplasia in a prepaid health care plan. Urology, suppl. 1, 38 13, 1991. 17. Morrison, A. S.:Risk factors for surgery for prostatic hypertrophy. Amer. J. Epidemiol., 136 974, 1992. 18. Daniell, H. W.: Larger prostatic adenomas in obese men with no associated increase in obstructive uropathy. J. Urol., 149 315, 1993. 19. Emrich, L. J., Priore, R. L., Murphy, G. P. and Brady, M. F.: Prognostic factors in patients with advanced stage prostate cancer. Cancer Res., 4 6 5173,1985. 20. Dewys, W. D., Begg, C., Lavin, P. T., Band, P. R., Bennett, J. M., Bertino, J. R., Cohen, M. H., Douglass, H. 0.. Jr., Engstrom, P. F., Ezdinli, E. Z., Horton, J., Johnson, G. J., Moertel, C. G., Oken, M. M., Perlia, C., Rosenbaum, C., Silverstein, M. N., Skeel, R. T., Sponzo, R. W. and Tormey, D. C.: Prognostic effect of weight loss prior to chemotherapy in cancer patients. Eastem Cooperative Oncology Group. Amer. J. Med., 6 9 491,1980. 21. Byar, D. P. and Corle, D. K.: Analysis of prognostic factors for prostatic cancer in the VACURG studies. In: Controlled Clinical Trials in Urologic Oncology. Edited by L. Denis, G. P. Murphy, G. R. Prout and F.Schroder. New York: Raven Press, pp. 147-169, 1984. 22. Daniell, H. W.: More stage A prostatic cancers, less surgery for benign hypertrophy in smokers. J . Urol., 149 68, 1993. 23. West, D. W.,Slattery, M. L., Robison, L. M., French, T. K and Mahoney, A. W.: Adult dietary intake and prostate cancer risk in Utah: a case-control study with special emphasis on aggressive tumors. Cancer Causes Control, 2: 85, 1991. 24. Mohle-Boetani, J. C., Grosser, S., Whittemore, A. S., Malec, M., Kampert, J . B. and Paffenbarger, R. S., Jr.: Body size, reproductive factors, and breast cancer survival. Prevent. Med., 17: 634,1988. 25. Senie, R.T., Rosen, P. P., Rhodes, P., Lesser, M. L. and f i n n e , D. W.: Obesity at diagnosis of breast carcinoma influences duration of disease-free survival. Ann. Intern. Med., 116 26, 1992. 26. Daniell, H.W.: Increased lymph node metastases a t mastectomy for breast cancer associated with host obesity, cigarette smoking, age, and large tumor size. Cancer, 62: 429,1988. 27. Daniell, H.W., Tam, E. and Filice, A.: Larger axillary metastases in obese women and smokers with breast cancer-an influence by host factors on early tumor behavior. Breast Cancer Res. Treat., 26: 193, 1993. 28. Michnovicz, J . J., Hershcopf, R. J., Naganurna, H., Bradlow, H. L. and Fishman, J.: Increased 2-hydroxylation of estradiol as a possible mechanism for the anti-estrogenic effect of cigarette smoking. New Engl. J. Med., 315 1305, 1986.
Vol. 155,220-225, January 1% Printed in U.S.A.
A BETTER PROGNOSIS FOR OBESE MEN WITH PROSTATE CANCER HARRY W. DANIELL* From the Department of Family h t i c e , University of California Medical School at Davis, Redding, California
ABSTRACT
Purpose: The possibility was investigated whether obesity is associated with a favorable prognosis in men with nonstage A prostatic cancer independent of other risk factors. Materials and Methods: A total of 235 men with nonstage A prostatic cancer diagnosed between 1983 and 1990 at a community hospital was evaluated by tumor grade and stage, degree of obesity, smoking habits, age, year of diagnosis and survival. Results: In comparison to tumors of less obese men those in men who were at least 10% overweight were more often stage B or C than D (odds ratio 2.3,95% confidence interval 1.4to 4.9). Smoking abstinence, patient age older than 70 years and low Gleason scores were also independently associated with less advanced tumors. The lower frequency of advanced stage tumors among those of obese men was most prominent for stage D2 disease in nonsmokers, progressively decreasing from 24% of 29 nonsmokers who had been underweight for many years before diagnosis to 8%of 24 who had been greatly overweight for many years (p <0.05). Tumor specific mortality was also lower in obese men and nonsmokers independent of tumor stage and grade with 5-year tumor specific mortality rates of 10%in obese nonsmokers, 27% in nonobese nonsmokers, 36% in obese smokers and 43% in nonobese smokers. These differences reflected a combination of obesity related differences in stage distribution and a more favorable prognosis for obese men among patients with tumors of similar stages. Conclusions: These observations suggest inhibition of prostate cancer growth and metastasis by increased endogenous estrogen, decreased endogenous testosterone or other systemic changes characteristic of obesity. Kcr WORDS:prostate, prostatic neoplasms, obesity, smoking, neoplasm metastasis
Invasion and metastasis by prostate cancer involve interplay between tumor aggression and host antitumor defenses. Contributions by these tumors to disease progression have commonly been evaluated by relating their characteristics to cancer stage at diagnosis, or rates of recurrence or tumor specific mortality. Interference with tumor progression by antitumor therapy has been evaluated by analysis of recurrence and mortality rates, or quantitation of tumor size or tumor generated biological markers. Associations between disease progression and host factors have rarely been reported although great differences in mortality from prostate cancer are recognized between populations in which the incidence of these tumors is apparently similar. Prostate cancer of more advanced stage and worse overall prognosis have been associated with younger age1 and African ancestry,2.3 relationships that are usually thought to have resulted from delayed diagnosis. In 1 large series prostate cancer prognosis was unrelated to patient age after controlling for tumor grade.4 Relationships between obesity and prostate cancer have not been well defined. Studies in different populations demonstrated that the frequency of these tumors in obese compared to nonobese men varies from a 4fold increases to less6v7 or no differen~e.~.S None of these studies analyzed cases by tumor stage. Both prospective series relating obesity to mortality from prostate cancer indicated a 20 to 35% increase in mortality in obese 11 These observations are compatible with a higher frequency of but more favorable prognosis for prostate cancer in overweight men. The growth and metastasis of most prostatic cancer are inhibited by estrogen supplementation or androgen deprivation. Since obese men characteristically have increased endogenous estrogen and decreased endogenous testosterone Accepted for publication July 21, 1995. * Requests for reprints: 2626 Edith Ave., Suite A, Redding, California 96001.
levels,12-14 the possibility of tumor inhibition induced by these altered sex hormone levels was evaluated by designing a study that included quantitation of obesity. This study examined the associations between several host factors, and tumor stage and tumor specific mortality in white men with nonstage A prostatic cancer.16 Obesity was independently associated with less advanced disease at diagnosis, as was smoking abstinence, older age and low tumor grade. Details of the associations between smoking, and tumor stage and prognosis have been previously described.16 "he present report analyzes data from this cohort, which relate obesity to tumor stage and tumor associated death. Fewer advanced tumors among those of obese men indicate less aggressive early disease progression unless there is an earlier diagnosis or tumor induced weight loss. A more favorable prognosis in obese men after controlling for tumor stage and grade, age and smoking habits would indicate less aggressive late disease progression. MATERIALS AND METHODS
The hospital records were reviewed of all 240 men with an initial diagnosis of nonstage A prostate cancer between July 1983 and December 1990 at Mercy Medical Center, 1 of 2 community hospitals serving a semirural, largely white population at the northern end of the central valley of California. Recorded data included year of diagnosis, patient age, height, weight, smoking habits, presenting signs and symptoms, prostate surgery performed, stage and Gleason grade at diagnosis. Survival data were obtained from personal physicians, hospital cancer registry and death certificates. Recent weight changes in stage D cases were calculated by comparing recorded hospital or office weights at the time of diagnosis with documented hospital or office weights recorded 1 to 4 years before cancer diagnosis. In 5 cases earlier weights were not
220
OBESITY AND PROSTATE CANCER PROGNOSIS
22 1
TABLE1. Characteristics of235 nonstage A prostate cancer patients analyzed by age at tumor diagnosis F’t. Aee at Diaenosis I
Younger Than 60
No. pts. Pt. characteristics (Yo): Smoking 1109 Recommended wt. or more 120% Recommended wt. or more 41, or More weight loss Pelvic lymphadenectomy Tumor characteristics: % Stage D1 ‘7r Stage D2 Av. grade* 1 5-Yr. tumor specific mortality * Not available for 3 tumors.
6044
-
0-99 100-109 110-119 120-129 130-139 140orGreater Totals
44 42 53 56 36
(21) (27) (31) (19) (5)
53
66
46
19
235
65 53 29 12 59
29 50 35 12 68
26 64 30 4 70
26 64 33 5 67
13 26 13 9 41
5 47 16 16 0
25 52 27 8 57
24 29 5.18 24
29 18 5.34 22
19 21 5.80 32
15 15 5.44 18
7 22 5.59 22
0 21 5.71 31
16 20 5.54 23
Total 70
(NO. pU.1
D1
D2
6 (3) 17(11) 24(14) 21 (7) 21 (3)
19 (9) 20(13) 7 (4) 18 ( 6 ) 14 (2)
44(15) 22(l4) 17 (9) 6 (2) 29 (4)
100 (48) 100 (65) 100 (58) 100 (34) 100 (14)
17 (41)
16 (37)
20 (46)
100 (235)
older), recommended weight (less than 109% versus 110% and greater) and smoking status. Probabilities of empirical results were determined by the chi-square test under the null hypothesis. Death certificates of 100 of the 104 men known to have died were reviewed. A total of 57 certificates listed prostate cancer as the primary cause of death and 7 included prostate cancer as a contributing cause. These 64 deaths were tabulated as prostate cancer deaths. The 4 men with an unknown cause o f death were tabulated as alive on the date preceding death. RESULTS
Table 1 shows tumor and patient characteristics. Older men were less often smokers or obese, and less likely to have undergone pelvic lymphadenectomy and, therefore, to have had stage D1 tumors. Factors unrelated to patient age included the percent of stage D2 tumors, tumor grade and 5-year tumor specific mortality. Table 2 lists the percent of men in each obesity group who presented with disease of each stage. Stage D2 tumors were less common among those of men in most weight categories of greater than 110% of recommended weight. Previously reported multivariate analysis demonstrated independent associations between stage D disease and lack of obesity (odds ratio 2.3,95%confidence interval 1.4 CO 4.9, p = 0.003),Gleason grade 7 to 10 (p = 0.001),patient age younger than 70 years (p = 0.008) and smoking (p = 0.015).’5 The association between stage D disease and younger age was largely but not completely due to the much higher frequency of pelvic lymphadenectomy in younger men (table 1). Gleason scores were unrelated to smoking habits or obesity (table 3). Table 4 compares weights recorded 1 to 4 years before diagnosis with those recorded at diagnosis, and indicates changes in weight categories during the interval between these observations. Ten patients (24%)with stage D2 disease lost more than 6% of body weight but only 4 had changed from above to below 110% recommended weight. Of the 6 patients whose category changed from obese to nonobese 3 had lost 7 or fewer pounds. Weight loss before diagnosis was TABLE3. Average Gleason scores for men with nonstage A prostate cancer Av. Score (No. tumors)* % Recommended Wt.
Smokers
0-99 100-109 110-119 12LL129 130 or Greater
Totals 47 (111)
Overall Values
.
34
% Tumor Stage (No. pts.)
C
Than80 Older
75-79
17
TABLE 2. Tumor stage and degree of obesity in men with nonstage A prostatic cancer B
70-74
~~
available and in 3 weights at diagnosis were confounded by the presence of edema. In these 3 patients dry weights measured l to 4 years before diagnosis were used in calculations. The smoking history of 184 men was obtained from the oftice or hospital records of urologists who required patients to complete forms detailing personal habits as part of the preoperative evaluation, and for most others from intake interviews by the nursing service, history documented by anesthesiologists or other consultants, or office records of attending urologists or personal physicians. Failing adequate data from these sources the history of tobacco use was obtained from 11 men or their families by the office staffs of their physicians. Height or weight of 5 men was unavailable, leaving 235 for analysis. Included were 3 patients with stage D2 disease for whom tumor grade was unavailable because the small amount of malignant tissue identified did not allow for grade determination (l),or biopsies were not done after presentation with multiple bony metastases and markedly elevated prostate specific antigen determinations (2). Men who had smoked a t least 10 cigarettes (1/2 pack) for a t least 5 years, including a portion of the 3 years before surgery, or who were currently documented smokers of cigarettes, a pipe or cigars were tabulated as smokers. More detailed analysis of smoking habit and its associations with tumor stage and tumor specific mortality has been previously reported.15 Obesity was quantitated by dividing recorded weight by the average weight for recorded height and medium habitus recommended in the 1983 Metropolitan Life Insurance height and weight tables, resulting in the percent recommended weight. Histological examination of surgical specimens was perfarmed by a group of pathologists who freely collaborated on difficult or unusual cases. Tumor grade was assigned by these pathologists after examining biopsy and staging tissues. If biopsy and prostatectomy tissues differed in tumor grade, the higher grade was tabulated. Stepwise logistic regression analysis was performed for associations between stage D tumors and Gleason score (2 to 6 versus 7 to lo), host-patient age (younger than 70 years versus 70 years and
1 Recommended Wt.
65-69
* Not available for 3 tumors.
6.1 (17) 5.5 (17) 4.8 (12) 5.0 (7) -6 5.5 (58)
Nonsmokers 5.1 (30) 5.4 (48) 5.5 (451 6.0 (26) 5.7 (25) 5.5 (174)
OBESITY AND PROSTATE CANCER PROGNOSIS
222
TAEW 4. Weight changes preceding diagnosis of stnge D prostatic cancer ~~
~
Sbge
D1
Stage D2
No. No. No. No. Smokers Nonsmokers Smokers Nonsmokers
0
25 1 0
17 1 2
10
24
14
1 27
0 0
0 2
3
3
0
8 1
7 0
1
18 2 1
0 15
4
3 6
0
0
0
2
2
0
0
4
0
0
1
0
1
0
1
2
12
No. pta. No data
2
Disquafified (edema)
Documented aerial wta. Change in % rearmmended
29 1
E
wt.:
Greater than 6% gain 44% Gain Less than 4% gain or loss
44% Loss Greater than 6%loss change in obesity categoly
049%
100109% 110-119% 120-12!4%
130%+
-
FIG. 1. Percent of nonstage A prostate cancers presentingat stage D2 in nonsmokers ouped by degree of patient obesity before tumor associated weight %ss. Values in parentheses indicate number of men.
110%
Greater than to leas than
'
PERCENT RECOMMENDEDWEIGHT 1 4 YEARS BEFORE DIAGNOSIS
before diagnosis:
Less than to greater than
0%
110%
Lesathantogreatertllm 120%
Greater tban to less than
unrelated to obesity in smokers or nonsmokers (data not shown). Of men with stage D2 disease those older than 75 years had more often lost weight recently than younger men (50% versus 19%, p <0.03). Table 5 lists the distribution of obesity categories in smokers and nonsmokers with stage D2 disease before and after tumor associated weight changes. No relationship is apparent in smokers between obesity and the presence of D2 disease. In nonsmokers a pattern of less frequent stage D2 disease in more obese men was suggested long before tumors were diagnosed (p
% Rscommended
wt.
and tumor stage. No relationship was suggested by the 5-year tumor specific mortality between younger and older men with similar smoking habits and obesity, an analysis handicapped by the small numbers in some age groups. Better &year tumor specific survival rates of obese men persisted in smokers and nonsmokers after correction for minor differences in age distribution. Tumor specific mortality was lower in obese than nonobese nonsmokers (p <0.01) and in obese nonsmokers than all others (10% versus 30%, p <0.001). Tumor specific survival rates for obese and nonobese smokers and nonsmokers with stage D disease, and all patients with nonstage A tumors are shown in figure 2. The longer survival of nonobese nonsmokers reflects a combination of lower mortality in those with tumors of identical stages (table 8) and a larger percent with less advanced tumors. Patients had received a wide variety of therapies, most of which were conventional, and administered with varying intensity and consistency. Some patients were lost to followup until death. No differences in antitumor therapy were noted in men of similar ages but with different smoking habits or degrees of obesity. DISCUSSION
These observations are compatible with the inhibition of prostate cancer growth, invasion and metastasis by factors associated with obesity, which are most prominent in nonsmokers and independent of other risk factors. Since many prostate cancers are uniquely sensitive to sex hormone manipulation, and to my knowledge a similar relationship has not been demonstrated in patients with other types of malignant tumors, the estrogen and testosterone changes characteristic of obesity seem likely to be responsible for these associations. The possibility that earlier diagnosis of prostate cancer in obese men contributes to the lower percent of advanced tumors at diagnosis seems remote. Rectal examinations are more difficult to perform in men with obese buttocks, which
TABLE5. Smoking habits and a'egree of obesity in men with stage 0 2 prostate cancer 1-4Yrs.Before Diagnosis. At Diagnosis 96 Smokers (No.)
96 Nonsmokers (No.)
% Smokers (No.)
0-99
35 (17)
100-109 110-119 120-129
13 (16) 8 (12)
24 (29) 20 (46) 14 (44) 10 (30) 8 (24)
39 (18) 18 (17) 15 (13)
38 18)
Greater than 130 50 (4) Not available for 3 smokers and 2 nonsmokers.
% Nonsmokers (No.)
29 (7) 60 ( 5 )
12 (25)
OBESITY AND PROSTATE CANCER PROGNOSIS T-LE p/r
223
6 . Smoking habits and degree of obesity in men with stage D l prostate cancer who underwent pelvic lymphadenectomy
Recommended Wt.
1-4 Yrs. Before Diagnosis* % Smokers (No.)
0-99 33 (6) 100-109 40 (10) 110-119 11 (9) 120-129 50 (4) Greater than 130 50 (2) * Not available for 2 smokers and 1 nonsmoker.
B Nonsmokers (No.)
36 (11) 24 (29) 15 (27) 25 (20) 33 (12)
TABLE7 . Men with nonstage A prostatic cancer diagnosed earlier (1983 to 1986) and later (1987 to 1990)
At Diagnosis 8 Smokers (No.)
9% Nonsmokers (No.)
50 (8) 50 (12) 0 (8) 33 (3) 50 (2)
42 (12) 24 (29) 15 (27) 25 (20) 33 (12)
indicate differences in the frequency or intensity of smoking, or undefined dietary, genetic, environmental or therapeutic differences, some of which may reflect the decades between NO. RS. (B stage D c a ) the previous and present observations, obesity related differ1983-1986 1987-1990 ences in the incidence of nonstage A prostate cancers andlor Obese nonsmokers 37 (24) 61 (23) statistical aberration. Nonobese nonsmokers 33 (48) 44 (34) Relationships between obesity and the incidence of prosObese smokers 16 (63) 8 (25) tate cancer are incompletely defined. Talamini et al reported Nonobese smokers 21 (62) 15 (47) on a population of men in northern Italy in which the inciTotals 107 (44) 128 (29.7) dence of prostate cancer in obese men was greater than 4 times that of men of the lowest quartile in percent body weight.5 A similar obesity related pattern of incidence in the frequently do not allow a thorough search for prostate nod- present population would make the data from this cohort ules. This pattern would be expected to delay the diagnosis of easily compatible with those reported by Snowden et all1 and palpable asymptomatic malignancies in some men. Obese Lew and Garfinkel.10 However, others noted a 20 to 35% men undergo less surgery for benign prostatic hypertro- increase in prostate cancer incidence6.7 or no increase*.g in phy16.'7 and have larger adenomas a t operation,la suggesting obese men. None of these incidence studies reported tumors later onset of obstructive symptoms in those with benign by stage, thereby failing to distinguish between potentially enlargement and longer delay in the diagnosis of some tu- lethal tumors and those of little clinical concern. The degree mors identified during the evaluation of symptomatic benign to which obesity related differences in prostate cancer incidisease. The possibility of inherently fewer aggressive malig- dence between these populations are due to differences in nancies among those of obese men as an explanation for the techniques of tumor identification or environmental, geolower percent of advanced tumors or lower tumor specific graphic and other factors largely remains to be investigated. mortality is not supported by the similarity of tumor grades No incidence studies have controlled for smoking habits, in the obese and nonobese men in this study. although our recent studies suggest a higher incidence of Tumor associated weight loss apparently contributed only stage A diseasezzand lower incidence of nonstage A disease15 slightly to the aforementioned associations although the fre- in smokers. Different risk factors for aggressive and nonagquency and degree of weight loss were similar to those in gressive prostate cancers are also suggested by the dietary larger studies. Of the patients with stage D2 disease in the studies reported by West et al.23 present series 28% had lost 4% or more recommended body The more favorable prognosis for obese than nonobese nonweight compared to 27% of 1,020 men with stage D2 dis- smokers with tumors of identical stages in the present series easels and 28% of 78 men with similarly advanced tumors in suggests that a t least some of the influence on disease proother series.2" In the only other study associating weight or gression associated with obesity may continue to function obesity with death rates in men with prostate cancer Byar even in the presence of conventional antitumor therapy. If and Corle reported fewer deaths from any cause in 80%of the this pattern is confirmed by future studies, it suggests that men with advanced prostate cancer who weighed more than conventional therapy may not offer the optimal therapeutic 59 kg. (130 pounds) than the 20% who weighed less. HOWever, no association was noted between body weight and combination of sex hormone supplementation and inhibition death attributable to prostate cancer, and they did not in- for many patients with advanced prostatic cancer, and that clude analysis by tumor stage or tumor associated weight optimal antitumor therapy may differ in men of different obesity status or smoking habits. loss.21 In the present series the deaths of almost all men with Both prospective studies that assessed mortality rates stage D disease who died during followup were due to prosfrom prostate cancer in men with different degrees of obesity demonstrated increased death rates from these tumors in tate cancer. These data suggest that obese nonsmokers with obese men. Lew and Garfinkel reported cause-specific mor- stage D disease may live longer if they maintain excess tality rates for many tumors in 336,442 men older than 30 weight despite increased death rates from cardiovascular Years enrolled for 12 years beginning in 1960.1° Men with a and other nonmalignant diseases associated with obesity. In recommended weight of 110 to 119%, 120 to 129%, 130 to this cohort deaths from all causes were less frequent in obese 139% and greater than 140% had age adjusted prostate can- than nonobese men with prostate cancer independent of tucer specific death rates of 0.9, 1.37, 1.33 and 1.29%, respec- mor stage. Appropriate additional studies will be of interest. Although estrogen and testosterone levels were not availtively, compared to 1% for those of 90 to 109%recommended able in the present study, data from other reports indicate weight. Smoking habits were not analyzed. Snowdon et a1 reported higher mortality from prostate cancer in 6,763 men that estrone production in men who are 30% more obese than weighing more than 130% followed for 20 years beginning in the recommended weight may approximate 120% of that in 1960.11 However, this obese group only included 13 deaths men of ideal weight,13,l4 an increase in estrogen influence from prostate cancer, the analysis did not include smoking equal to approximately that of0.12 mg. diethylstilbestrol per habits and men with a recommended weight of 110 to 129% day, and that testosterone production in these obese men had a death rate from prostate cancer that was almost iden- may have decreased to between 85% and 90% of that of tical to that of less obese men. Differences between t h e men of similar age,12Bi3changes of a lower magnitude than Present results and those of the 2 preVious Studies may those induced by conventional hormonal therapy. These o h -
OBESITY AND PROSTATE CANCER PROGNOSIS
224
TABLE8. 5-Year cumulative mortality in men with nonstage A prostate cancer according to tumor stage, smoking habits and obesity before tumor associated weight lossx All Cause Mortality
Tumor SpecSc Mortality Tumor
% Smokers
% Nonsmokers
% Smokers
stage
(No. men)
(No. men)
(No. men)
Obese
Nonobese 90 (10)
02
26 (8) 16 (15) 42 (33) *Weight at diagnosis in 8 men.
BandC E D
A
Obese
Nonobese
Obese
Nonobese
Obese
65 119) 27 (12) 8 (46) 25 (77)
53 110) 8 (13) 5 (75) 10 (98)
90 (10) 26 ( 8 ) 41 (15) 52 (33)
86 (7) 25 (4) 14 (16) 34 (27)
68 (19) 27 (12) 24 (46) 35 (77)
61 (10) 27 (13) 15 (75) 21 (98)
B
PROSTATE CANCERS
1
2
3
4
5
6
PROSTATE CANCERS
7
1
2
YEARS
4
5
6
7
7
MEN REMAINING
-
10
7
7
3
3
3
2
27
25
22
21
16
12
9
18
15
12
11
10
6
6
3
*----*33
29
25
24
20
13
11
7
25
23
22
19
14
12
5
3
98
97
92
84
66
49
25
15
29
24
17
17
12
10
9
6
-----a 77
65
57
54
44
34
23
17
-11
O----O
3 YEARS
MEN REMAINING *-----O
(No. men)
Nonobese
86 (71 25 (4) 14 (16) 34 (27)
D1
% Nonsmokers
FIG.2. A, tumor specificsurvival of obese and nonobese smokers, and nonsmokers with stage D prostate cancer. Obesity status preceded tumor associated weight loss or gain. B, tumor speciSc survival of obese and nonobese smokers, and nonsmokers with nonstage A prostate cancer. sity related differences in hormone production were established in men younger than 60 years and may be greater or less than differences in those in older age groups, which include most men with prostate malignancies. Others have suggested modification of the behavior of breast cancers, another group of sex hormone sensitive malignancies, by the obesity associated changes in these hormones.=.% In women with breast cancer obesity is associated with more advanced disease at diagnosis,26 larger axillary metastasis27 and a less favorable prognosis after controlling for tumor stage and other fa~tors.24~25 In the present study the greater improvement in prognosis associated with obesity in nonsmokers suggests that the mechanism by which this improvement is effected may be neutralized in smokers. This pattern may reflect the increased 2-hydroxylation of estradiol demonstrated in smokers by Michnovicz et al,m which has been proposed as the mechanism responsible for the antiestrogen influence promoting premature menopause, osteoporosis and other estrogen deficiency diseases in women who use tobacco. The prominent stage migration after 1986 in the present series and those of others has been thought to occur on the basis of earlier tumor diagnosis in later years resulting from more widespread use of prostate specific antigen determinations and a more aggressive pursuit of early prostate cancer diagnosis by physicians and patients. However, in the present study the more frequent obesity and less frequent smokingby patients in whom the disease was diagnosed after
1986 were responsible for approximately an eighth of the stage migration in this cohort. If smoking habits and degree of obesity had remained the same after 1986 as before that date and the percent of advanced tumors in each smokingobesity category had retained the improvement noted during the later interval, the percent of those with stage D disease diagnosed after 1986 would have been 31.5% rather that 29.7%, compared to 44.9% before 1986. Similarly, stage migration contributed little to the association between obesity and less advanced tumor stage. Of these men 28.7% who were and 45.1% who were not obese presented with stage D disease. If the percent of men diagnosed after 1986 had been identical in obesity and smoking status to that before 1986 and the lower percent of advanced tumors had been the same, 29.2% of obese and 45.7%of nonobese men would have presented with stage D disease. CONCLUSIONS
Many studies examining mortality from prostate cancer have shown that a 5-year followup is insufficient to demonstrate differences that are clearly present during longer intervals due to the slow progression of the majority of prostate malignancies. Although much of the improved 5-year tumor specific mortality rate associated with obesity in the present cohort was due to tumor related differences in stage distnbution (table 7), the prognosis of obese nonsmokers with stage D disease was clearly more favorable than that of
OBESITY AND PROSTATE CANCER PROGNOSIS
nonobese nonsmokers with similar tumors (fig. 2, A). Longer followup should clarify the relative importance of obesity related differences in stage distribution and tumor behavior after diagnosis to the improved survival of obese men with prostate cancer. The present technique for establishing smoking habits, largely obtained from chart review, includes considerable potential error, which has been discussed previously.15 The similar technique for identifying the height and weight of each patient in the present study also includes the potential for error but data similar to those tabulated had usually been documented by different recorders in multiple office and hospital chart locations, and often during different periods of hospitalization. These weaknesses in study design seem unlikely to have modified the described associations. Prospective studies using more conventional techniques for quantitation of tobacco use and comparison of the prognosis of men who stop smoking or gain weight after diagnosis to that of men who do not will be of great interest. Dr. T. Joseph Sheehan performed the multivariate analysis, Billie White prepared the graphics and Phyllis Watkins and Valerie McRae assisted in patient followup. REFERENCES
1. Hamson, G. S. M.: The prognosis of prostatic cancer in the younger man. Brit. J. Urol., 5 5 315, 1983. 2. Brawn, P. N., Johnson, E. H., Kuhl, D. L., Riggs, M. W., Speights, V. O., Johnson, C. F., 111, Pandya, P. P., Lind, M. L., and Bell, N. F.: Stage at presentation and survival of white and black patients with prostate carcinoma. Cancer, 71: 2569, 1993. 3. Polednak, A. P. and Flannery, J . T.: Black versus white racial differences in clinical stage a t diagnosis and treatment of prostatic cancer in Connecticut. Cancer, 7 0 2152, 1992. Damber, J.-E., Jonsson, H. and Lenner, P.: Patient 4. Gronberg, H., age as a prognostic factor in prostate cancer. J. Urol., 152:892, 1994. 5. Talamini, R., La Vecchia, C., Decarli, A., Negri, E. and Franceschi, S.: Nutrition, social factors and prostatic cancer in a Northern Italian population. Brit. J . Cancer, 53: 817,1986. 6. Nomura, A., Heilbrun, L. K. and Stemmermann, G. N.: Body mass index as a predictor of cancer in men. J. Natl. Cancer Inst., 74:319, 1985. 7. Mills, P. K., Beeson, W. L., Phillips, R. L. and Fraser, G. E.: Cohort study of diet, lifestyle, and prostate cancer in Adventist men. Cancer, 64:598, 1989. 8. Kolonel, L. N., Yoshizawa, C. N. and Hankin, J . H.: Diet and prostatic cancer: a case-control study in Hawaii. h e r . J. Epidemiol., 127: 999, 1988. 9. Graham, S.,Haughey, B., Marshall, J., Priore, R., Byers, T., Rzepka, T,, Mettlin, C. and Pontes, J. E.: Diet in the epidemiology of carcinoma of the prostate gland. J . Natl. Cancer Inst., 7 0 687,1983. 10. h w , E . A. and Garfinkel, L.: Variations in mortality by weight among 750,000 men and women. J. Chron. Dis., 3 2 563,1979. 11. Snowdon, D. A,, Phillips, R. L. and Choi, W.: Diet, obesity and risk of fatal prostate cancer. Amer. J . Epidemiol., 120 244, 1984. 12. Zumoff, B., Strain, G. W., Miller, L. K., Rosner, W., Senie, R..
225
Seres, D. S. and Rosenfeld, R. S.: Plasma free and non-sexhormone-binding-globulin-bound testosterone are decreased in obese men in proportion to their degree of obesity. J. Clin. Endocriol. Metab., 71: 929,1990. 13. Simon, D., Preziosi, P., Barrett-Connor, E., Roger, M., Saint-Paul, M., Nahoul, K. and Papoz, L.: The influence of a g h g on plasma sex hormones in men: the Telecom Study. Amer. J. Epidemiol., 136 783, 1992. 14. Kley, H. K,Deselaers, T., Peerenboom, H. and Kruskemper, H. L.: Enhanced conversion of androstenedione to estrogens in obese males. J . Clin. Endocrinol. Metab., 51: 1128, 1980. 15. Daniell, H. W.: A worse prognosis for smokers with prostate cancer. J . Urol., 154:153, 1995. 16. Sidney, S.,Quesenberry, C., Jr., Sadler, M. C., Lydick, E. G., Guess, H. A. and Cattolica, E. V.: Risk factors for surgically treated benign prostatic hyperplasia in a prepaid health care plan. Urology, suppl. 1, 38 13, 1991. 17. Morrison, A. S.:Risk factors for surgery for prostatic hypertrophy. Amer. J. Epidemiol., 136 974, 1992. 18. Daniell, H. W.: Larger prostatic adenomas in obese men with no associated increase in obstructive uropathy. J. Urol., 149 315, 1993. 19. Emrich, L. J., Priore, R. L., Murphy, G. P. and Brady, M. F.: Prognostic factors in patients with advanced stage prostate cancer. Cancer Res., 4 6 5173,1985. 20. Dewys, W. D., Begg, C., Lavin, P. T., Band, P. R., Bennett, J. M., Bertino, J. R., Cohen, M. H., Douglass, H. 0.. Jr., Engstrom, P. F., Ezdinli, E. Z., Horton, J., Johnson, G. J., Moertel, C. G., Oken, M. M., Perlia, C., Rosenbaum, C., Silverstein, M. N., Skeel, R. T., Sponzo, R. W. and Tormey, D. C.: Prognostic effect of weight loss prior to chemotherapy in cancer patients. Eastem Cooperative Oncology Group. Amer. J. Med., 6 9 491,1980. 21. Byar, D. P. and Corle, D. K.: Analysis of prognostic factors for prostatic cancer in the VACURG studies. In: Controlled Clinical Trials in Urologic Oncology. Edited by L. Denis, G. P. Murphy, G. R. Prout and F.Schroder. New York: Raven Press, pp. 147-169, 1984. 22. Daniell, H. W.: More stage A prostatic cancers, less surgery for benign hypertrophy in smokers. J . Urol., 149 68, 1993. 23. West, D. W.,Slattery, M. L., Robison, L. M., French, T. K and Mahoney, A. W.: Adult dietary intake and prostate cancer risk in Utah: a case-control study with special emphasis on aggressive tumors. Cancer Causes Control, 2: 85, 1991. 24. Mohle-Boetani, J. C., Grosser, S., Whittemore, A. S., Malec, M., Kampert, J . B. and Paffenbarger, R. S., Jr.: Body size, reproductive factors, and breast cancer survival. Prevent. Med., 17: 634,1988. 25. Senie, R.T., Rosen, P. P., Rhodes, P., Lesser, M. L. and f i n n e , D. W.: Obesity at diagnosis of breast carcinoma influences duration of disease-free survival. Ann. Intern. Med., 116 26, 1992. 26. Daniell, H.W.: Increased lymph node metastases a t mastectomy for breast cancer associated with host obesity, cigarette smoking, age, and large tumor size. Cancer, 62: 429,1988. 27. Daniell, H.W., Tam, E. and Filice, A.: Larger axillary metastases in obese women and smokers with breast cancer-an influence by host factors on early tumor behavior. Breast Cancer Res. Treat., 26: 193, 1993. 28. Michnovicz, J . J., Hershcopf, R. J., Naganurna, H., Bradlow, H. L. and Fishman, J.: Increased 2-hydroxylation of estradiol as a possible mechanism for the anti-estrogenic effect of cigarette smoking. New Engl. J. Med., 315 1305, 1986.