Smoking and adverse outcomes at radical prostatectomy
Background: Multiple large epidemiologic studies have examined the relationship between smoking and prostate cancer incidence and mortality only to arrive at contradictory results. In this series, we studied the effect of smoking on pathologic outcomes and biochemical recurrence in a cohort of men undergoing radical prostatectomy. Methods: We identified 630 men who underwent radical prostatectomy between 1989 and 2005 who had detailed smoking histories. There were 321 smokers and 309 nonsmokers. Pathologic outcomes included prostate weight, volume of cancer, volume of high grade cancer, margin status, seminal vesicle involvement, extraprostatic extension, perineural invasion, angiolymphatic invasion, and the presence of nodal metastasis. Biochemical recurrence was defined as a postoperative PSA ≥ 0.1 ng/ml. Univariate analysis and multivariate linear and Cox regression were used to study the impact of smoking on these outcomes. Results: The volume of cancer (2.54 vs. 2.16 ml, P = 0.016) and the volume of high grade cancer (0.58 vs. 0.28 ml, P = 0.004) were greater in smokers compared with nonsmokers. Smoking independently predicted greater volumes of cancer and high grade cancer in multivariate analysis. Heavy smokers (≥20 pack-year history) had a greater risk of biochemical recurrence on univariate survival analysis. Smoking also predicted a greater risk of biochemical recurrence on Cox regression, the magnitude of which was approximately 1% per pack-year smoked. Conclusions: Smoking is associated with adverse pathologic features and a higher risk of biochemical recurrence in men undergoing radical prostatectomy. If confirmed by additional studies, smoking history may need to be included into risk assessment models.
Smoking and adverse outcomes at radical prostatectomy
Tin C. Ngo, M.D.
*, J. Joy Lee, M.D.
, James D. Brooks, M.D.
, Rosie Nolley, B.A.
Michelle Ferrari, R.N.
, Joseph C. Presti Jr., M.D.
Department of Urology, Stanford University School of Medicine, Stanford, CA 94305, USA
Received 25 May 2011; received in revised form 23 June 2011; accepted 23 June 2011
Background: Multiple large epidemiologic studies have examined the relationship between smoking and prostate cancer incidence and
mortality only to arrive at contradictory results. In this series, we studied the effect of smoking on pathologic outcomes and biochemical
recurrence in a cohort of men undergoing radical prostatectomy.
Methods: We identiﬁed 630 men who underwent radical prostatectomy between 1989 and 2005 who had detailed smoking histories.
There were 321 smokers and 309 nonsmokers. Pathologic outcomes included prostate weight, volume of cancer, volume of high grade
cancer, margin status, seminal vesicle involvement, extraprostatic extension, perineural invasion, angiolymphatic invasion, and the presence
of nodal metastasis. Biochemical recurrence was deﬁned as a postoperative PSA ⱖ 0.1 ng/ml. Univariate analysis and multivariate linear
and Cox regression were used to study the impact of smoking on these outcomes.
Results: The volume of cancer (2.54 vs. 2.16 ml, P ⫽ 0.016) and the volume of high grade cancer (0.58 vs. 0.28 ml, P ⫽ 0.004) were greater
in smokers compared with nonsmokers. Smoking independently predicted greater volumes of cancer and high grade cancer in multivariate analysis.
Heavy smokers (ⱖ20 pack-year history) had a greater risk of biochemical recurrence on univariate survival analysis. Smoking also predicted a
greater risk of biochemical recurrence on Cox regression, the magnitude of which was approximately 1% per pack-year smoked.
Conclusions: Smoking is associated with adverse pathologic features and a higher risk of biochemical recurrence in men undergoing
radical prostatectomy. If conﬁrmed by additional studies, smoking history may need to be included into risk assessment models. © 2013
Elsevier Inc. All rights reserved.
Keywords: Prostate cancer; Smoking; Tobacco; Cancer volume; Biochemical recurrence
Although cigarette smoking has been implicated in many
cancers and is the leading cause of cancer mortality, its role in
prostate cancer remains ill-deﬁned . Numerous epidemio-
logic studies have attempted to link smoking with prostate
cancer incidence [2–5] and mortality [6–9] but have yielded
contradictory results. One explanation for these divergent ﬁnd-
ings may be inherent biases and problems with misclassiﬁca-
tion of exposure and outcomes associated with large survey-
based cohort studies. Additionally, some of these studies
utilized cohorts that originated prior to the PSA era and thus
their results may not be generalized to contemporary patients.
Nevertheless, because prostate cancer accounts for approxi-
mately 25% of all newly diagnosed malignancies in US men
 and the prevalence of cigarette smoking is 21.6% ,
clarifying the inﬂuence of smoking on prostate cancer has
important public health and clinical implications.
The Stanford Radical Prostatectomy Database (SRPD), a
prospectively maintained repository of all radical prostatecto-
mies (RP) performed at our institution, contains information on
patient demographics, medical history, pathologic features at
the time of surgery, and clinical outcomes. One of its strengths
is that it has detailed morphometric data on the total cancer
volume and high grade (Gleason pattern 4 and 5) cancer
volume as determined by McNeal et al. . Because infor-
mation on smoking history for these patients was also avail-
able, this cohort served as a convenient population in which to
study the relationship between cigarette smoking and prostate
cancer in patients treated by RP. In this study, we analyzed
both pathologic outcomes and biochemical recurrence (BCR).
2. Materials and methods
2.1. Study population
With institutional review board approval (Internal Re-
view Board Approval: Approved by the Stanford University
School of Medicine IRB, Protocol 11714), we reviewed the
* Corresponding author. Tel.: ⫹1-408-813-7049; fax: ⫹1-408-503-0050.
E-mail address: email@example.com (T.C. Ngo).
Urologic Oncology: Seminars and Original Investigations 31 (2013) 749 –754
1078-1439/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
SRPD and identiﬁed 739 patients who had detailed mor-
phometric information on cancer volume and high grade
cancer volume who also had clinical follow-up. Of these,
630 had information on smoking history. The RP were
performed from 1989 to 2005 by multiple surgeons. There
were 309 nonsmokers and 321 smokers. Smoking history
was obtained from self-reported patient intake forms. These
data were reviewed with the patient by the treating surgeon
at the initial consultation. Our analysis was restricted to
those with detailed information about their pathologic and
clinical outcomes. Pathologic outcomes included prostate
weight, volume of cancer, volume of high grade cancer,
margin status, seminal vesicle involvement (SVI), ex-
traprostatic extension (EPE), perineural invasion (PNI), an-
giolymphatic invasion (ALI), and pelvic lymph node in-
volvement (PLNI). Clinical outcomes included BCR as
deﬁned by a postoperative PSA ⱖ 0.1 ng/ml.
2.2. Pathologic examination and morphometric analysis
Immediately after RP, fresh specimen were weighed and
ﬁxed in formalin. They were then sectioned axially at 3 mm
intervals. These sections were then cut at 5
m and exam-
ined microscopically by a single pathologist (JM), a recog-
nized expert in prostate cancer pathology. Areas of cancer on
each slide were traced manually. The volume of cancer was
then calculated by taking the sum of the area of cancer on each
slide multiplied by its thickness using previously described
software . The volume of cancer demonstrating Gleason
pattern 4 or 5 was deemed the volume of high grade cancer.
2.3. Statistical analysis
Age, PSA levels, prostate weight, volume of cancer, and
volume of high grade cancer were continuous variables and
reported as medians with an associated interquartile range.
The presence of positive margins, SVI, EPE, PNI, ALI,
PLNI, and BCR were categorical variables and reported as
proportions. Differences between medians were tested using
the nonparametric Wilcoxon rank sums test and the
Kruskal-Wallis one-way analysis of variance, depending on
the number of groups being compared. Differences between
proportions were tested using Pearson’s
test. Based upon
prior studies [7,14] we suspected a dose-response effect of
smoking on pathologic and clinical outcomes and, thus,
dichotomized our smoking group into light smokers and
heavy smokers using a cutoff value of 20 pack-year. Light
smokers were thus deﬁned as those with a ⬍20 pack-year
history while heavy smokers were with a ⱖ20 pack-year
history. This cut point was selected based on a prior anal-
yses done in a large scale cohort study looking at the effect
of smoking on prostate cancer mortality .
Univariate analysis was performed to study the above
pathologic outcomes as a function of smoking history (non-
smoker vs. smoker; nonsmoker vs. light smoker vs. heavy
smoker). Multivariate linear regression was performed to
study the effect of age, PSA, prostate weight, Gleason score,
and smoking on the volume of cancer and the volume of
high grade cancer. In the multivariate analysis, smoking was
deﬁned as a continuous variable based on pack-year.
Survival analysis using the Kaplan-Meier product limit
estimator was used to compare the rates of BCR between
smokers (all smokers, light smokers, and heavy smokers)
and nonsmokers. Differences between groups were tested
with the log-rank test. Multivariate survival analysis with
Cox proportional hazards modeling was used to determine
the effect of age, PSA, Gleason score, and clinical T stage,
volume of cancer, volume of high grade cancer, and smok-
ing (as a continuous and dichotomous variable) on the risk
of BCR. Because of a higher likelihood of occult metastatic
disease in those with node positive disease, these patients
were excluded from the survival analysis of BCR (n ⫽ 31).
All statistical testing was 2-sided, and a P ⱕ 0.05 was
considered statistically signiﬁcant. Covariates that were not
normally distributed were log transformed to facilitate sta-
tistical analyses when needed. JMP Statistical Discovery
Software (SAS Institute, Cary, NC) was used for all statis-
The demographics of patients in this series are shown in
Table 1. Of the 630 patients included in this study, 309
never smoked while 321 had a history of smoking. Of the
smokers, 197 had detailed pack-year information. The me-
Nonsmokers 309 (49%)
Smokers 321 (51%)
Median PSA (ng/ml)
Pack-years among smokers
Year of prostatectomy
Before 1989 7 (1.1%)
1989–1993 174 (27.6%)
1994–1998 310 (49.2%)
1999–2005 139 (22.1%)
Median follow-up (mo)
Biochemical 125 (19.8%)
Local 11 (1.7%)
Distant 29 (4.6%)
* Recurrences are mutually exclusive.
750 T.C. Ngo et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 749 –754
dian number of pack-years among the smokers was 25
(IQR, 12–40). The distribution of prostatectomies in this
cohort over time was reported using intervals previously
described, with the vast majority of the cases occurring in
the PSA era . The median follow up in this study was
71.6 months (IQR, 43.5–101.4).
The association between smoking and various pathologic
and clinical outcomes were evaluated with univariate anal-
yses (Table 2). When we treated smoking history as a
dichotomous variable (smoking vs. nonsmoking), only the
differences in volume of cancer (2.54 vs. 2. 16 ml, P ⫽
0.016) and volume of high grade cancer (0.58 vs. 0.28 ml,
P ⫽ 0.004) between the two groups were statistically sig-
niﬁcant. When smoking history was treated as a categorical
variable (nonsmoker vs. light smoker vs. heavy smoker),
volume of cancer and volume of high grade cancer also had
statistically signiﬁcant differences between the three
groups. There were no statistically signiﬁcant associations
between rates of positive margins, SV involvement, EPE,
PNI, ALI, or PLNI and smoking.
Having identiﬁed the association between smoking and
volume of cancer and volume of high grade cancer, we
performed a multivariate linear regression to examine the
inﬂuence of smoking and other covariates on these patho-
logic outcomes (Table 3). In these models, smoking was
coded as a continuous variable (pack-year). As one would
expect PSA had a highly signiﬁcant, positive
in our models, suggesting that it closely correlates with
cancer volume. Additionally, pack-year smoked had a sta-
tistically signiﬁcant positive
coefﬁcient in both models,
suggesting that it is directly proportional to volume of
cancer and volume of high grade cancer independent of the
other covariates in the model. Controlling for age, prostate
weight, and PSA, each unit increase in pack-year was as-
sociated with a 0.03 ml increase in volume of cancer and
0.02 ml increase in volume of high grade cancer.
The inﬂuence of smoking on BCR was studied with
univariate survival analysis using the Kaplan-Meier product
limit estimator. When nonsmokers were compared with all
smokers, there was no statistically signiﬁcant difference
between the Kaplan-Meier curves for the 2 groups (log-rank
P ⫽ 0.93). Likewise, when nonsmokers were compared
with light smokers, there was no signiﬁcant difference be-
tween those 2 groups (log-rank P ⫽ 0.20). Since nonsmok-
ers and light smokers were similar in terms of BCR, we
combined them and compared the resulting group with the
heavy smoker group and found a statistically signiﬁcant
difference in the rates of BCR (log-rank P ⫽ 0.03) (Fig. 1).
To conﬁrm this ﬁnding and to assess for other covariates
that may contribute to the risk for BCR, multivariate anal-
yses using Cox proportional hazards models were per-
formed (Table 4). One model used smoking as a dichoto-
mous variable (0 –19 pack-year vs. ⱖ20 pack-year) and
another used smoking as a continuous variable. In both
models, the inﬂuence of PSA, volume of high grade cancer,
and smoking history had hazard ratios for BCR that were
greater than unity and were statistically signiﬁcant (see
Table 4). Cancer volume itself was not signiﬁcant in either
model, likely because it correlates so closely with PSA that its
effect on the hazard ratio is dampened in the models. In
separate analyses, when PSA level was dropped from our
models, volume of cancer became statistically signiﬁcant (not
shown). These data suggest that PSA, volume of high grade
cancer, and smoking were independent predictors of BCR.
Univariate analysis of pathologic and clinical outcomes as a function of
Parameter Nonsmoker Light
Median 62.5 61.8 63.3 0.33
IQR 57.1–67.1 57.0–67.0 58.9–67.3
Median 7.59 6.6 7.27 0.55
IQR 4.88–11.1 5.2–10.7 4.6–9.6
Prostate weight (g)
Median 46 41.5 45 0.08
IQR 38–59.5 34.8–52 36–60
Volume of cancer (ml)
Median 2.16 1.96 2.98 0.05
IQR 1.07–4.6 1.34–3.58 1.22–5.37
Volume of high grade
Median 0.28 0.41 0.69 0.01
IQR 0–1.08 0.07–1.18 0.1–1.89
Positive margin (%) 14.9% 15.20% 23.10% 0.11
8.4% 6.10% 11.60% 0.38
26.0% 21.50% 31.50% 0.29
Perineural invasion (%) 80.3% 81.10% 88% 0.31
16.0% 14.80% 22.90% 0.25
Nodal involvement (%) 5.3% 4.70% 5.50% 0.97
* Light smoker is deﬁned as having a ⬍20 pack-year history of smoking.
Heavy is ⱖ20 pack-year.
** High Grade is deﬁned as containing Gleason pattern 4 or 5.
Multivariate linear regression of volume of cancer and high grade cancer
Parameter Coefﬁcient (
) SE 95% CI P
Volume of cancer
Age 0.0203 0.0282 ⫺0.0361–0.0767 0.47
Prostate weight ⫺0.0081 0.0085 ⫺0.0251–0.0089 0.34
Log(PSA) 5.8808 0.5680 4.7448–7.0168 ⬍0.0001
Pack-year 0.0319 0.0082 0.0155–0.0483 0.0001
Volume of high grade cancer*
Age 0.0227 0.0176 ⫺0.0125–0.0579 0.2
Prostate weight ⫺0.0086 0.0053 ⫺0.0192–0.0020 0.1
Log(PSA) 3.2588 0.3541 2.5506–3.9670 ⬍0.0001
Pack-year 0.0151 0.0051 0.0049–0.0253 0.003
* High grade is deﬁned as containing Gleason pattern 4 or 5.
751T.C. Ngo et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 749 –754
A clear relationship between smoking and prostate can-
cer has been elusive. While several large scale epidemio-
logic studies spanning the last 2 decades have hinted at an
association between smoking and prostate cancer mortality,
conﬂicting studies have also been published. A large scale
retrospective cohort that included over 55,102 men span-
ning the years 1975 to 2002 showed that those with a ⱖ20
pack-year history of smoking had a higher risk of death
from prostate cancer (RR ⫽ 2.38) . Similarly, the Health
Professionals Follow-Up Study, a cohort of 47,781 men
between 1986 and 1994, found that men who smoked more
than 15 pack-year had a greater risk of fatal prostate cancer
(RR ⫽ 2.06) . Yet, the analysis of the Physician’s Health
Study cohort of 22,071 men showed no statistically signif-
icant increase in the risk of prostate cancer death . Al-
though these large cohort studies have the beneﬁt of statis-
Fig. 1. Kaplan Meier curves for heavy smokers (20 or more pack-years) and non-heavy smokers (less than 20 pack years). (Color version of ﬁgure is available
Cox proportional hazards analysis of biochemical recurrence
Parameter Coefﬁcient (
) SE Hazard ratio 95% CI P
Smoking as a dichotomous variable
Age ⫺0.0122 0.0139 0.9878 0.9614–1.0154 0.38
Volume of cancer ⫺0.0119 0.0287 0.9881 0.9306–1.0420 0.67
Log (PSA) 1.3915 0.3166 4.0208 2.1617–7.4728 ⬍0.0001
Volume of high grade cancer* 0.2878 0.0424 1.3335 1.2296–1.4531 ⬍0.0001
Heavy smoker** 0.2385 0.1018 1.2693 1.0346–1.5438 0.02
Smoking as a continuous variable
Age ⫺0.0122 0.0141 0.9878 0.9611–1.0157 0.39
Volume of cancer ⫺0.0104 0.0290 0.9897 0.9314–1.0440 0.72
Log (PSA) 1.4043 0.3187 4.0727 2.1767–7.5909 0.0001
Volume of high grade cancer* 0.2826 0.0428 1.3266 1.2226–1.4472 ⬍0.0001
Pack-year 0.0070 0.0031 1.0070 1.0001–
* High grade is deﬁned as containing Gleason pattern 4 or 5.
** Heavy smoker is deﬁned as having a ⱖ20 pack-year history of smoking.
752 T.C. Ngo et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 749 –754
tical power, they suffer from problems with selection and
recall bias, inaccurate assignment of exposures and out-
comes, and patient attrition. Additionally, these cohorts
originated before the PSA era, and conclusions based their
data may not generalize to contemporary patients. Our study
represents a small subset of contemporary prostate cancer
patients, speciﬁcally those treated by RP. Precise histo-
pathologic characterization allowed us to better study the
inﬂuence of smoking on cancer and biologic endpoints.
We have shown that cigarette smoking independently
and in a dose-responsive manner predicts (1) larger volumes
of cancer, (2) larger volumes of high grade cancer, and (3)
a greater risk of BCR in men undergoing RP for treatment
of prostate cancer.
In univariate analysis, the volume of cancer was signif-
icantly different between nonsmokers and smokers. When
we further dichotomized the smoking group into light smok-
ers and heavy smokers, the difference remained statistically
signiﬁcant. These results were conﬁrmed on multivariate
analysis showing that pack-year smoked predicted a higher
volume of cancer in a highly signiﬁcant manner. Similar
results were obtained when the relationship between vol-
ume of high grade cancer and smoking history was ex-
amined. These ﬁndings were also conﬁrmed on multivar-
Smoking is also associated with a greater risk of BCR on
both univariate and multivariate analysis. The ﬁnding that
heavy smokers had a higher risk of BCR compared with
light smokers and nonsmokers suggests a dose-dependent
relationship. This relationship was also seen within the Cox
proportional hazards models in which greater hazard ratios
were observed for heavy smoking status (HR ⫽ 1.269) and
pack-years smoked (HR ⫽ 1.007 per pack-year).
It is also notable that we did not detect an association
between smoking and adverse pathologic outcomes other
than cancer volume and high grade cancer volume. Al-
though 1 report has shown an association between smoking
and EPE, it was limited to young men ⬍ 55 years old and,
thus, may be susceptible to selection bias . In contrast,
our study population did not restrict patients by age.
Because the association between cancer volume and
BCR has been well characterized , the question thus
becomes whether smoking itself directly imparts a greater
risk of BCR or indirectly through its association with higher
cancer volumes. The results from our study suggest that it
does both since, in our multivariate models looking at co-
variates that affect cancer volume, high grade cancer vol-
ume, and BCR, smoking acts as an independent variable.
Nevertheless, a recent study from the Shared Equal Ac-
cess Regional Cancer Hospital (SEARCH) database found
no differences in BCR between smokers and nonsmokers
. We believe that the authors of that study were unable
to detect a difference in the risk of BCR because their
dataset did not include detailed smoking history in terms of
pack-year. As we described, when survival analysis was
performed on our cohort using smoking as a dichotomous
variable (nonsmoker vs. smoker), no difference could be
observed in the risk of BCR. However, when heavy smok-
ers, which we deﬁned as those with at least a 20 pack-year
history of smoking, were compared with nonsmokers, light
smokers, and the combination of the 2, the survival curves
showed that heavy smokers suffered a statistically signiﬁ-
cant greater risk of BCR. These results, along with the fact
that pack-year behaves as an independent variable in our
Cox proportional hazards model, argue that the effect of
smoking on BCR is dose-dependent, and may have gone
undetected in their study if all smokers were pooled into one
The biologic basis for these observations has yet been
clearly deﬁned, though there are several potential mecha-
nisms to explain how smoking may impact the biology of
prostate cancer. Cadmium containing fertilizers are used in
the cultivation of tobacco and several studies have shown a
link between cadmium exposure and prostate cancer initia-
tion and progression in rats and an association between
cadmium levels and advanced prostate cancer in humans
[19 –21]. Additionally, exposure to well known carcinogens
found in cigarette smoke including polycyclic aromatic hy-
drocarbons, heterocyclic aromatic amines, and nitrosamines
may promote progression to higher grade cancers by induc-
ing mutations in regulatory genes such as p53 [22–24].
Another potential explanation lies in the observation that
men who smoke have higher circulating levels of androgens
and therefore have an internal hormonal milieu that pro-
motes cancer growth [25–28]. The exact molecular mecha-
nism by which smoking results in larger cancers and more
aggressive cancer phenotypes warrants further study.
This study has some important limitations. It is a single
institutional experience at an academic referral center and
thus may not be applicable to other settings in which patient
demographics and pretreatment clinical features differ. Fur-
thermore, we did not have data on potential confounding
variables such as race, BMI, statin use, second-hand smoke
exposure, or current and former smoking status. Also, we
lacked pack-year information for approximately one-third
of the smokers, which may lead to selection bias. Never-
theless, even if we excluded the analysis that included
smoking in terms of pack-years, there would still be the
ﬁnding that smoking is associated with higher volumes of
cancer and high grade cancer, 2 independent predictors of
For men undergoing radical prostatectomy for prostate
cancer, a history of smoking is associated with adverse
pathologic features and a higher risk of BCR. If conﬁrmed
by additional studies with larger cohorts, smoking history
may need to be included into risk assessment models. The
biologic basis for this ﬁnding remains unclear and warrants
753T.C. Ngo et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 749 –754
further investigation. Whether smoking cessation after sur-
gery inﬂuences oncologic outcome or not is unknown.
 Hickey K, Do KA, Green A. Smoking and prostate cancer. Epidemiol
 Cerhan JR, Torner JC, Lynch CF, et al. Association of smoking, body
mass, and physical activity with risk of prostate cancer in the Iowa
65⫹ Rural Health Study (United States). Cancer Causes Control
 Hiatt RA, Armstrong MA, Klatsky AL, et al. Alcohol consumption,
smoking, and other risk factors and prostate cancer in a large health
plan cohort in California (United States). Cancer Causes Control
 Lumey LH, Pittman B, Zang EA, et al. Cigarette smoking and
prostate cancer: No relation with 6 measures of lifetime smoking
habits in a large case-control study among U.S. whites. Prostate
 Rohan TE, Hislop TG, Howe GR, et al. Cigarette smoking and risk of
prostate cancer: A population-based case-control study in Ontario and
British Columbia, Canada. Eur J Cancer Prev 1997;6:382– 8.
 Rohrmann S, Genkinger JM, Burke A, et al. Smoking and risk of fatal
prostate cancer in a prospective U.S. study. Urology 2007;69:721–5.
 Giovannucci E, Rimm EB, Ascherio A, et al. Smoking and risk of
total and fatal prostate cancer in United States health professionals.
Cancer Epidemiol Biomarkers Prev 1999;8:277– 82.
 Lotufo PA, Lee IM, Ajani UA, et al. Cigarette smoking and risk of
prostate cancer in the physicians’ health study (United States). Int J
Cancer 2000;87:141– 4.
 Adami HO, Bergstrom R, Engholm G, et al. A prospective study of
smoking and risk of prostate cancer. Int J Cancer 1996;67:764 –8.
 Jemal A, Siegel R, Ward E, et al. Cancer statistics 2009. CA Cancer
J Clin 2009;59:225– 49.
 Cokkinides V, Bandi P, McMahon, C, et al. Tobacco control in the
United States—recent progress and opportunities. CA Cancer J Clin
 McNeal JE, Bostwick DG, Kindrachuk RA, et al. Patterns of pro-
gression in prostate cancer. Lancet 1986;1:60 –3.
 McNeal JE, Villers AA, Redwine, EA, et al. Capsular penetration in
prostate cancer. Signiﬁcance for natural history and treatment. Am J
Surg Pathol 1990;14:240 –7.
 Rodriguez C, Tatham LM, Thun MJ, et al. Smoking and fatal prostate
cancer in a large cohort of adult men. Am J Epidemiol 1997;145:
 Stamey TA, Caldwell M, McNeal JE, et al. The prostate speciﬁc
antigen era in the United States is over for prostate cancer: What
happened in the last 20 years? J Urol 2004;172:1297–301.
 Roberts WW, Platz EA, Walsh PC. Association of cigarette smoking
with extraprostatic prostate cancer in young men. J Urol 2003;169:
 Chung BI, Tarin TV, Ferrari M, et al. Comparison of prostate cancer
tumor volume and percent cancer in prediction of biochemical recur-
rence and cancer speciﬁc survival. Urol Oncol 2011;29:314 –8.
 Moreira DM, Antonelli JA, Presti JC Jr. et al. Association of cigarette
smoking with interval to biochemical recurrence after radical prosta-
tectomy: Results from the SEARCH database. Urology 2010;76:
 Waalkes MP, Anver MR, Diwan BA. Chronic toxic and carcinogenic
effects of oral cadmium in the Noble (NBL/Cr) rat: Induction of
neoplastic and proliferative lesions of the adrenal, kidney, prostate,
and testes. J Toxicol Environ Health A 1999;58:199 –214.
 Waalkes MP, Rehm S. Cadmium and prostate cancer. J Toxicol
Environ Health, 1994;43:251– 69.
 Chen YC, Pu YS, Wu HC, et al. Cadmium burden and the risk and
phenotype of prostate cancer. BMC Cancer 2009;9:429.
 Nock NL, Liu X, Cicek MS, et al. Polymorphisms in polycyclic
aromatic hydrocarbon metabolism and conjugation genes, interac-
tions with smoking and prostate cancer risk. Cancer Epidemiol Bio-
markers Prev 2006;15:756 –61.
 Pour PM. Prostatic cancer induced in MRC rats by N-nitrosobis(2-
oxopropyl)-amine and N-nitrosobis(2-hydroxypropyl)amine. Carci-
nogenesis 1983;4:49 –55.
 Quinones LA, Irarrazabal CE, Rojas CR, et al. Joint effect among
p53, CYP1A1, GSTM1 polymorphism combinations and smoking on
prostate cancer risk: An exploratory genotype– environment interac-
tion study. Asian J Androl 2006;8:349 –55.
 Shiels MS, Rohrmann S, Menke A, et al. Association of cigarette
smoking, alcohol consumption, and physical activity with sex steroid
hormone levels in US men. Cancer Causes Control 2009;20:877– 86.
 Field AE, Colditz GA, Willett WC, et al. The relation of smoking,
age, relative weight, and dietary intake to serum adrenal steroids, sex
hormones, and sex hormone-binding globulin in middle-aged men.
J Clin Endocrinol Metab 1994;79:1310 –6.
 Dai WS, Gutai JP, Kuller LH, et al. Cigarette smoking and serum sex
hormones in men. Am J Epidemiol 1988;128:796 –805.
 Saylor PJ, Smith MR. Metabolic complications of androgen depriva-
tion therapy for prostate cancer. J Urol 2009;181:1998 –2006.
754 T.C. Ngo et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 749 –754