Accuracy of Self-reported Smoking Exposure Among Bladder Cancer Patients Undergoing Surveillance at a Tertiary Referral Center

Abstract

Background

Smoking is a risk factor for developing bladder cancer (BCa). Even though continued exposure after diagnosis may adversely affect prognosis, patients may be reluctant to disclose to their physicians that they are currently smoking, leading to inaccurate reporting of exposure and missed opportunities to deliver smoking-cessation advice and treatment in the context of cancer care.

Objective

We examined the extent of misclassification of recent smoking exposure among patients undergoing BCa surveillance.

Design, setting, and participants

A consecutive sample of 145 patients with a self-reported smoking history and prior initial diagnosis of BCa was recruited from a tertiary referral urology clinic.

Outcome measurements and statistical analysis

Patients were asked if they had smoked a cigarette or used nicotine replacement therapy (NRT) within the past week and whether they lived with a smoker. At the same visit, we collected urine under a biospecimen protocol. We used urinary cotinine, the primary metabolite of nicotine, as an objective biomarker of recent smoking exposure. Nine patients whose urine could not be interpreted for cotinine were excluded. We calculated the smoking status misreporting rate by comparing biochemically verified smoking status (≥31.5 ng/ml vs <31.5 ng/ml) against self-reported current smoking status (yes vs no) while considering recent NRT use.

Results and limitations

Overall, 11% (15 of 136) of patients had cotinine values consistent with current smoking. Of these 15 patients, 7 reported being former smokers, resulting in a 47% misclassification rate. However, three of the seven patients who denied smoking in the past week were currently using NRT. Excluding NRT users, the misclassification rate was 33%.

Conclusions

Future studies investigating the impact of postdiagnosis nicotine exposure on BCa outcomes should use biochemical verification combined with self-reported smoking exposure to classify patients accurately.

Patient summary

Bladder cancer patients may misreport smoking exposure, thereby missing opportunities for smoking cessation.

1. Introduction

Cigarette smoking is an established risk factor for developing bladder cancer (BCa), and continued exposure after diagnosis may adversely affect prognosis by increasing the risk of recurrence and progression as well as compromising intravesical therapy response [1–4]. Moreover, emerging data suggest that nicotine exposure after diagnosis may enhance tumor growth and metastasis [5–9]. An accurate assessment of smoking exposure is essential for reducing misclassification in studies on the impact of smoking on BCa outcomes and for referring patients to smoking-cessation programs. Identifying BCa patients who are current smokers is especially important in light of the recent findings that smokers are less likely to adhere to American Urological Association guidelines regarding surveillance cystoscopies [10].

All prior studies investigating associations between smoking and BCa outcomes have relied on self-report to capture patient smoking history and recent exposure [1,3,4,11,12]. Studies in smokers without cancer suggest that 32% of smokers require biochemical assessment to identify true tobacco use [13]; the limited studies among cancer patients suggest that misreporting rates are variable and that self-reported smoking may be inaccurate in up to 55% of patients [14–18]. Most recently, Morales et al [17] found that patients who had a smoking-related cancer (ie, lung cancer) were more likely to misrepresent tobacco use than those patients who had breast cancer or prostate cancer (PCa). Among 77 lung cancer patients, the misreporting rate was 60% versus 23% in 79 PCa patients. Even when patients were aware of secondary biochemical verification of their smoking status, Alberg et al found a 39% misreporting rate in a cohort of 108 head and neck cancer patients [18]. To our knowledge, no prior study has evaluated the accuracy of self-reported cigarette smoking among BCa patients. This population is noteworthy because BCa patient awareness of smoking as a risk factor for their disease is limited [19–22]. The purpose of this study was to describe the extent of misclassification of recent smoking exposure among BCa patients undergoing surveillance at a tertiary referral center.

2. Material and methods

Over 4 mo in 2013, we recruited 145 consecutive BCa patients who had a smoking history, were being treated at Memorial Sloan Kettering Cancer Center, and had consented to an existing specimen-collection protocol. We used urinary cotinine, the primary metabolite of nicotine, as an objective biomarker to determine recent smoking exposure [23]. The treating physician asked patients if they had smoked a cigarette within the past 7 d (yes vs no), had used nicotine replacement therapy (NRT) products within the past 7 d (yes vs no), and currently lived with an active smoker (yes vs no). At the same visit, urine was collected for biochemical assessment of cotinine using established methods of liquid chromatography coupled with tandem mass spectrometry and atmospheric pressure chemical ionization. Patients were blinded to the purpose of the study. We could not assess the urine of nine patients for cotinine because of interference, leaving a total of 136 patients for this analysis. Continuous cotinine values were categorized into three groups—≥31.5 ng/ml, 0.5–31.4 ng/ml, and <0.5 ng/ml which—represented current, passive, and no recent smoking exposure, respectively [24,25]. We derived our cut-off points from clinical pharmacology studies that were designed to determine the optimal cut-off point of cotinine to discriminate active versus passive versus no recent nicotine exposure. We dichotomized patients into cotinine-positive (≥31.5 ng/ml) or cotinine-negative (<31.5 ng/ml) subgroups to represent current smoking status (yes vs no). We calculated sensitivity, specificity, false-positive, and false-negative rates from a 2 x 2 classification table comparing biochemically verified smoking status (≥31.5 ng/ml vs <31.5 ng/ml) against self-reported current smoking status (yes vs no). We abstracted additional patient and disease characteristics, such as a more detailed smoking history, from electronic medical records. All statistical analyses were performed using SAS version 9.4 software (SAS Institute, Cary, NC, USA).

3. Results

Table 1 presents the demographic, clinical, and smoking characteristics of the 136 BCa patients who had a smoking history. Patients were predominantly male and white; mean age at sample collection was 73 yr. The year patients were initially diagnosed with BCa ranged from 1994 to 2013. The majority of patients presented with early stage disease (<pT2; 95.6%) and high-grade tumors (68.4%). The median number of months since diagnosis was 31. The median number of cigarettes smoked per day and lifetime duration of smoking was 1 pack and 25 yr, respectively. Among self-reported former smokers, the median number of years since quitting was 27, with an interquartile range of 13–38 yr.

Table 1.

Demographic, clinical, and smoking characteristics of bladder cancer (BCa) patients with a smoking history currently under surveillance for BCa recurrence (n = 136)

Characteristics	Total no. (%)
Age in yr, median (IQR)	73 (66–78)
Gender
Male	115 (85)
Female	21 (15)
Race
Caucasian	129 (94.9)
Black	3 (2.2)
Asian	2 (1.5)
Other	1 (0.7)
Missing	1 (0.7)
Year of initial BCa diagnosis	1994–2013
Months since initial diagnosis, median (IQR)	31 (19–73)
Stage
<pT2	130 (95.6)
≥pT2	6 (4.4)
Grade
High	93 (68.4)
Low	38 (27.9)
Missing	5 (3.7)
Average no. cigarettes smoked per day, median (IQR)	20 (20–40)
Lifetime duration of smoking in years, median (IQR)	25 (15–35)
Years since quitting, median (IQR)*	27 (13–38)

BCa = bladder cancer; IQR = interquartile range.

^*Among 128 former smokers.

Table 2 compares self-reported smoking status with biochemically verified urinary cotinine levels. Overall, 11% (15 of 136) of patients had cotinine values consistent with current smoking (range: 118.3–2047.8 ng/ml). Of these 15 patients, 7 reported being a former smoker, resulting in a misclassification rate of 47%. Three of the seven patients who reported being a former smoker said they currently used NRT. All the patients classified as cotinine negative reported that they were former smokers (121 of 121, 100%). Only 5 (4%) of these patients had values consistent with passive smoking exposure (range: 0.59–4.65 ng/ml). Self-reported smoking status was 53% sensitive and 100% specific. There were no false positives, but there were seven false negatives. The positive predictive value was 100%, and the negative predictive value was 94%.

Table 2.

2 × 2 Table comparing self-reported smoking status to urinary cotinine level

Smoked in past 7 d	Cotinine positive ≥31.5 ng/ml	Cotinine negative <31.5 ng/ml	Total
Smoked in past 7 d, yes	8	0	8
Smoked in past 7 d, no	7	121	128
Total	15	121	136

4. Discussion

Recent cigarette smoking exposure can be captured through self-report or biomarkers. In our cohort of BCa patients under surveillance at a tertiary referral center, we found that 11% of patients were classified as cotinine positive, 3% of whom reported currently using NRT. The misclassification rate among current smokers in our cohort of BCa patients was 47%. Our findings highlight the potential need for biochemical verification of recent smoking exposure among BCa patients to appropriately identify those who could be referred to smoking-cessation clinics and to reduce misclassification in future research studies.

To our knowledge, this pilot study is the first to assess the accuracy of self-reported smoking exposure among BCa patients. Our findings are in line with prior studies among patients with other smoking-related cancers that have observed misclassification rates as high as 55.8% [14–18]. Morales found that lung cancer patients who claimed that they had recently quit smoking were more likely than nonrecent quitters to misrepresent their current tobacco use. Although our sample size was too small to make such comparisons, the most recent quitter in our analysis was among those who misreported their smoking exposure.

Notably, several patients in our cohort who were classified as cotinine positive admitted to currently using NRT. A growing body of cell culture and experimental animal studies demonstrate that nicotine, although not a known carcinogen, promotes cell proliferation, angiogenesis, and epithelial-to-mesenchymal transition through nicotinic acetylcholine receptors found in the bladder, leading to enhanced tumor growth and metastasis [5–9]. It is also possible for smoking exposure to compromise response to intravesical treatment for BCa because constituents of cigarette smoke are potent inhibitors of cytokine production [26], particularly those involved in bacillus Calmette-Guérin treatment response, including interleukin 2, tumor necrosis factor (TNF), and interferon-γ (IFN-γ). Thus, future research studies designed to estimate the effect of recent nicotine exposure on BCa outcomes should measure cotinine to accurately classify their patients.

We were surprised to find a relatively low proportion of patients with a smoking history who had cotinine values consistent with current smoking. Prior studies of BCa survivors found that approximately 25% of patients with a smoking history continue to smoke after diagnosis (based on self-report) [1,19,27]. Reasons for this discrepancy are unclear but may relate to differences in how current smoking was defined across studies [28]. In addition, the availability of a smoking-cessation program for cancer patients at our institution and the encouragement of smoking cessation by our physicians during their frequent follow-up visits may have contributed to the low smoking prevalence. Indeed, Bassett et al found that advice from their urologist was the most-cited reason for cessation among BCa patients who quit smoking after diagnosis [19].

Strengths of this investigation include the relatively large cohort of BCa patients with urine specimens who were blinded to the purpose of the study as well as data about NRT use and second-hand smoking exposure collected the same day. Limitations are acknowledged. Our study cohort consisted of consecutive BCa patients undergoing surveillance at a single tertiary referral center who have been living with the disease for many years, which limits generalizability. The low smoking prevalence and lack of variability in time since quitting prevented us from identifying factors associated with misreporting. Interference was noted in 9 of the 145 collected urine specimens because of coelution of urinary constituents with a retention time similar to urinary cotinine during gas chromatography. Repeated extraction steps lead to loss of urinary cotinine along with suspected sources of interference, including urinary contaminants or other prescription medications [29]. Assuming, however, that all 9 patients with urinary cotinine interference were in fact cotinine positive and self-reported smoking status accurately, the misreporting rate would still be 29% (7 of 24). Finally, we did not consistently assess other sources of nicotine exposure, such as electronic cigarette [30] or cigar use among those who denied being active smokers, which should be considered in future studies.

5. Conclusions

Patient misreporting represents a lost chance for tobacco-cessation counseling in the context of cancer care. Prospective studies that investigate the impact of postdiagnosis nicotine exposure on BCa outcomes should use biochemical verification in combination with self-report to classify patients accurately.