Impact of Tobacco Smoking on Outcomes of Radiotherapy: A Narrative Review

The carcinogenic role of tobacco smoking is well recognized, but the detrimental effects of continued smoking after a cancer diagnosis have been underestimated. Radiotherapy is among the main treatment modalities for cancer. We reviewed the literature data concerning the impact of tobacco smoking on treatment outcomes in radiotherapy-managed patients with various malignancies. Most of the analyzed studies demonstrated the detrimental effect of smoking on overall survival, tumor control, quality of life, treatment toxicity, and the incidence of second primary malignancies. Healthcare professionals should use the cancer diagnosis and treatment as a teachable moment and recommend their patients to immediately cease smoking. Wherever possible, cancer patients should undergo an intensive smoking-cessation program, including behavioral and pharmacologic therapy.


Introduction
Tobacco smoking is a well-documented causative factor for at least 13 malignancies [1]. More than 60% of cancer patients are current or former smokers [2]. Numerous studies and meta-analyses have demonstrated that continuing smoking after a cancer diagnosis negatively impacts survival and quality of life (QoL), while increasing treatment toxicity and the risk of secondary malignancies [3][4][5]. Radiotherapy (RT) is one of the main treatment modalities in cancer, contributing to around 40% of cures [6]. Approximately half of cancer patients will receive RT during the course of their disease [7].
The purpose of this article is to review the impact of tobacco smoking on RT outcomes in various malignancies. The examples provided here should be considered illustrative, as no comprehensive analysis of the literature pertaining to the topic was attempted.

Materials and Methods
The narrative review was done according to SANRA guidelines (https://researchintegrityjournal.biomedcentral.com/articles/10.1186/s41073-019-0064-8; accessed 15 September 2021). In October 2021, we performed a search using PubMed, Scopus, and Google Scholar. We used following search query: "radiotherapy AND (tobacco OR cigarette) AND smoking AND cancer". Additionally, articles were found in references of included articles and by Google search using "smoking during radiotherapy" during following months. Table 1 presents the summary of included studies. We included studies reporting (1) the number of active smokers during RT or chemoradiotherapy (CHRT) and non-smokers or past smokers as a control group, and (2) the outcomes for both group with regard to at least one of the specific endpoints: overall survival (OS) ( Table 2), progression-free-survival (PFS), distant-free-survival (DFS), locoregional control (LRC), distant control ( Table 3), risk of secondary primary cancer (SPC) ( Table 4), and treatment toxicity and complications (Table 5).     Articles were excluded if authors did not report any of the data mentioned in the including criteria. If multiple publications of the same cohort were available, only the latest publication was selected. Studies in languages other than English were excluded from this narrative review.

Lung Cancer
Of all malignancies, lung cancer is the most greatly associated with tobacco smoking. This association is stronger for squamous and small cell lung cancer (SCLC) than for adenocarcinoma and large cell carcinoma [65].

Non-Small Cell Lung Cancer
The impact of tobacco smoking on OS in non-small cell lung cancer (NSCLC) patients managed with RT or CHRT is unclear. In a study by Tsao et al., including 1370 advanced NSCLC patients (497 of whom were administered CHRT), the median OS in never, former, and current smokers in the CHRT cohort was similar (1.5, 1.5, and 1.4 years, respectively) [8]. There were also no significant differences in response rates between these groups (63%, 59%, and 50%, respectively). In a study by Nguyen et al., including stage II/III NSCLC patients administered postoperative RT, smokers, compared to non-smokers, had worse five-year local control (LC): 70% versus 90%, respectively (p = 0.001 and LRC: 52% versus 77% (p = 0.0006) [9]. In a study including stage I patients treated with stereotactic body radiation therapy (SBRT), no statistical differences were observed in terms of three-year LC rate (52% versus 56%) and OS between patients with a history of smoking and never smokers [10]. However, SBRT-induced toxicity occurred in 14.5% of current or past smokers and none of the never smokers. In another study including NSCLC patients managed with SBRT, smoking cessation was associated with improved two-year OS compared to continued smoking (78% versus 55%, respectively; p = 0.014) [11]. Finally, in the study by Miller et al., smoking status was not associated with distant failure risk [12].
Fox et al. showed inferior OS in smoking compared to non-smoking stage I/II NSCLC patients managed with RT or CHRT (41% versus 56%, respectively; p = 0.01) [13]. In a recent Russian study of Sheikh et al., the adjusted median OS and five-year OS in smokers and quitters were 4.8 years versus 6.6 years (p = 0.001) and 49% versus 61%, respectively (p = 0.001), and the five-year PFS was 44% versus 54%, respectively (p = 0.004) [14].
In a study by Rades et al., smoking during RT was associated with worse two-year LCR (34% for smokers versus 59% for non-smokers; p < 0.001), but this was not translated into a significant OS difference [15]. Paradoxically, a few studies reported higher rates of radiation-induced pulmonary (RP) toxicity in non-smokers than in smokers [16][17][18].
These observations were contrary to the Monson et al. study, which showed a 23% and 0% prevalence of RP toxicity in smoking and non-smoking patients, respectively (p < 0.01) [19]. In a study by Sarihan et al., the prevalence of infections was highly associated with tobacco exposure (p = 0.001) [20]. In turn, frequent infections were associated with inferior OS (median 9 versus 13 months, respectively; p = 0.042).
Kawaguchi et al. reported more than five-fold higher relative risk (RR) of SPC in stage III NSCLC smokers treated with CHRT compared to the general population [21]. The risk in those who quit smoking was not significantly higher than the risk in the general population.

Small Cell Lung Cancer
The first report on the impact of smoking during CHRT or chemotherapy (CHT) in SCLC was published by Johnston et al. in 1980 [22]. The median OS for smokers, quitters at diagnosis, and quitters before the diagnosis, was 47, 52, and 70 weeks, respectively (p < 0.04). In the study of Videtic et al., including limited-disease SCLC patients managed with CHRT, the median OS in former and continuing smokers was 18.0 months and 13.6 months, respectively (p = 0.002), and five-year OS was 8.9% and 4.0%, respectively (p = 0.0017) [23]. Patients who continued smoking during treatment and experienced a treatment-related break had the poorest OS. In another study, the RR of second lung cancer in continued smokers and quitters was 9.1 and 21, respectively, compared with general population [24]. In a study by Kawahara et al., including SCLC patients managed with CHRT or CHT, SPC occurred in 33% of patients who continued smoking, in 10% of patients who quit smoking, and in none of the never-smokers [25]. The risk of developing SPC in patients who continued smoking was significantly higher than in non-smokers (p = 0.03).

Head and Neck Cancer
Around 70% to 80% of head and neck cancers (HNCs) are linked to tobacco smoking [66]. In a study by Browman et al., HNC patients who continued to smoke tobacco during RT or CHRT had a lower complete response rate (45% versus 74%, respectively; p = 0.008) and two-year OS (39% versus 66%, respectively; p = 0.002) than those who abstained from smoking [26]. In a subsequent analysis of this study in another cohort of HNC patients, the median OS in abstainers and very light smokers was 42 months compared to 29 months in light, moderate, or heavy smokers (p = 0.07) [27]. However, smoking during RT was not an independent negative predictive factor, as opposed to smoking status at baseline. Somewhat surprisingly, in the study of Kawakita et al., including HNC patients managed with RT or CHRT, non-smokers had significantly worse OS than light smokers (p = 0.016), which raised doubts regarding the allocation of patient groups [28]. In a prospective study from Denmark, smokers had inferior OS and LRC compared to non-smokers (p = 0.03 and 0.02, respectively) [29].
In a study by Fortin et al., including HNC patients managed with definitive or postoperative CHRT, the OS was impaired in current and former smokers versus never-smokers (p = 0.000001) [30]. LC was lower in current compared to former smokers (p = 0.0001). Additionally, the LC curve of current smokers continued to decrease over time, contrary to the curve of former smokers, which reached a plateau after three years. For never, former, and current smokers, the five-year LC rates were 75%, 80%, and 67%, respectively (p < 0.0000001).
The detrimental impact of tobacco exposure was also reported in oropharyngeal cancer patients participating in phase III clinical trials of Radiation Therapy Oncology Group (RTOG) 9003 and 0129 [31]. Patients with an exposure higher than 10 PYs at diagnosis managed with definitive RT had an approximate doubled risk of death compared to those with up to 10 PYs (p = 0.001), corresponding to an absolute 30% difference in five-year OS.
Smoking remained an important predictor of OS and PFS after adjustment for tumor p16 status and other significant prognostic factors (p < 0.001 and <0.001, respectively). In the RTOG 0129 study, tobacco exposure negatively impacted OS in patients administered CHRT (p < 0.05). In the study cohorts of RTOG 0129 and 9003 trials, the five-year locoregional failure rate was higher in patients smoking more than 10 PYs versus up to 10 PYs at diagnosis (29% and 12%, respectively; p = 0.001 and 48% versus 26%, respectively; p = 0.01). Additionally, smoking exposure at diagnosis was the only significant factor associated with a higher risk of developing an SPC. In a study including 1875 human papillomaviruspositive oropharyngeal cancer patients managed with CHRT or RT, current smokers had inferior OS, PFS and DFS than non-smokers [32]. Finally, in a meta-analysis comprising 24 studies and 6332 HNC patients, smoking during or after RT was associated with a nearly doubled risk of death (p < 0.0001) and more than doubled risk of locoregional failure (p = 0.0005) [33].
Chen et al. showed superior three-year PFS in HNC patients managed with definitive CHRT who quit smoking after diagnosis, compared to continuing smokers (61% versus 34%, respectively; p = 0.05) [34]. Continuing smokers also had more grade ≥3 acute toxicities (p = 0.01) and a higher probability of gastrostomy or tracheostomy (p = 0.03). In a study by Egestad et al., including HNC patients administered postoperative RT or CHRT, smoking status was associated with increased fatigue, pain, speech disturbances, mouth opening problems, and poorer cognitive function [35]. Likewise, in a Danish study, smokers, compared to non-smokers, showed worse cognitive function, nausea/vomiting, dyspnea, appetite loss, diarrhea, and weight loss (p < 0.05) [36]. Finally, in a study by Silveira et al., continued smoking decreased patients' QoL by negatively affecting burden (p = 0.003), mental health (p = 0.03), and fatigue (p = 0.028) [37]. In a study including 10,381 nasopharyngeal cancer patients administered intensity-modulated RT or CHRT, current smokers had higher risk of death and recurrence compared to never-smokers (p = 0.003 and 0.027, respectively), whereas the risk of metastasis was higher in the former and current smokers (p = 0.031 and 0.019, respectively) than in never-smokers [39].
In the study by Al-Mamgani et al. in T1a glottis cancer patients, continued smoking smokers, compared to posttreatment quitters, had worse LC (81% versus 94; p = 0.001, OS (36% versus 70%; p < 0.001) and increased incidence of SPC (21% versus 12%; p = 0.003) [40]. Additionally, quitters showed better voice quality during two-year follow-up (p < 0.001 for all time points). In another study including patients with T1 glottis cancer, tobacco smoking was the single most important factor influencing the rate of RT-related complications (p = 0.031 after Bonferroni correction) [41]. The adjusted ten-year complication rates were 28% for continuing smokers, 11% for those who quit after RT, 14% for those who quit before RT, and 26% for never smokers. In a study by Porock et al., smoking HNC patients were more likely to develop oral mucositis (p = 0.03), and its severity was associated with tobacco exposure (p = 0.008) [42]. In a study by Rugg et al., mucositis lasted longest in patients who smoked during RT (23.4 weeks in current smokers, 18.3 weeks in those who temporarily abstained, 13.6 weeks in those who stopped before RT, and 12.4 weeks in never smokers) [43]. In a recent meta-analysis, HNC patients who ceased smoking developed less late toxicities, particularly osteoradionecrosis, than those who continued smoking [33]. In another study including HNC patients treated with RT, surgery, or a combination, current smokers had a significantly higher risk of death and SPC than never or former smokers [44].

Breast Cancer
In a cancer registry-based study including 10,676 breast cancer (BC) patients administered postoperative RT, the mortality was significantly higher for current or former-smokers than in never-smokers [45]. Additionally, smoking increased the risk of RT-induced SPC. In a study including 1029 BC patients subjected to lumpectomy and RT, those who continued smoking at the time of RT had a 20% 15-year risk of SPC compared to 16% for those who were non-smokers (p = 0.07) [46]. The 15-year risk of developing lung cancer after RT was 0.26% for non-smokers, 4.7% for former smokers, and 6.0% for smokers at the time of RT (p = 0.06). In a study by Sharp et al., including BC patients receiving postoperative RT, tobacco smoking was an independent factor associated with severe acute skin reactions (p = 0.031) [47]. Patients with such reactions showed a worse QoL (p = 0.007) and higher levels of pain and insomnia (p < 0.05). In another study, smoking and RT showed more than an additive effect on the risk of myocardial infarction (p = 0.039) [48]. In a meta-analysis including 71 studies using adjuvant RT, smoking was associated with worse post-reconstruction outcomes, such as higher failure rate of capsular contracture reconstruction, and other major complications. [49].

Prostate Cancer
Two studies demonstrated a negative effect of tobacco smoking in prostate cancer (PC) managed with brachytherapy [50,51]. The study by Taira et al. showed a more than doubled RR of death in former versus never-smokers (p = 0.007) and more than four-fold RR in current versus never-smokers (p < 0.001) [50]. In the second study, the HR for OS was 1.4 (p = 0.017) for never versus former smokers and 2.9 (p < 0.001) for never-versus current smokers [51]. In another study including patients managed with external RT, HR for OS was 1.72 (p = 0.08) and for distant control 5.24 (p = 0.003) in current versus neversmokers [52]. In a study including 61 PC patients managed with definitive external beam RT, six-year OS in non-smokers and smokers was 26% and 11%, respectively (p = 0.009) [53]. Alsadius et al. reported a higher prevalence of abdominal cramps (p = 0.004), defecation urgency (p < 0.001), sensations of bowel not completely emptied after defecation (p = 0.003), and sudden emptying of bowels into clothing without forewarning (p = 0.003) among smoking compared to non-smoking PC patients managed with external-beam RT [54]. In a study by Boorjian et al., the HR of secondary bladder cancer was 3.65 in smoking patients who received RT [55]. Finally, in the meta-analysis by Foerster et al., including 22,546 PC patients managed with prostatectomy or RT, the HR of biochemical recurrence in current and former-smokers receiving RT was 1.50 (1.20-1.88) and 1.10 (0.94-1.28), respectively [56].

Cervical Cancer
In a study by Mayadev et al., two-year pelvic control and DFS in heavy smoking cervical cancer women administered concurrent CHRT (>21 PY) were inferior to nonsmokers (p = 0.004 and 0.011, respectively) [57]. In a study by Eifel et al., including International Federation of Gynecology and Obstetrics stage I and II patients managed with RT, heavy smoking was the strongest independent predictor of all complications (p < 0.0005) [58]. The most striking smoking-related adverse effect were small bowel complications (for smokers of ≥1 pack per day: p < 0.0005). In turn, in the study by Fyles et al., smoking status was not associated with PFS and LC [59].

Other Malignancies
In a large Chinese series, smoking esophageal cancer patients managed with CHRT had a five-year OS of 32% compared to 51% in never-smokers (p = 0.01) [60]. There was no significant difference in OS between former and current smokers. Heavy smokers (>47.5 PY) had a poorer five-year OS than light smokers (16% versus 38%, respectively; p < 0.001). In a study by Lerman et al. including anal cancer patients administered RT or CHRT, smoking was the only independent factor significantly related to PFS (p = 0.013) [61]. In a study by Leeuven et al., patients who smoked more than 10 PY after a diagnosis of Hodgkin lymphoma experienced a six-fold elevated risk of secondary lung cancer compared with patients who smoked less than one PY (p = 0.03; adjusted for RT dose) [62]. In a study by Peppone et al. including BC, genitourinary cancer, lung cancer, gastrointestinal cancer, and other malignancies managed with RT, CHT, or both, the mean total symptom burden during treatment was significantly greater in smoking than in non-smoking patients (46% versus 41%; p = 0.048) [63]. Patients who quit smoking before treatment had a total symptom burden comparable to non-smokers. In another study including BC, HNC, and anorectal cancers, current smokers had higher rates of RT-induced skin reactions than former smokers [64]. Smoking was associated with higher RTOG scores (p < 0.0001), erythema meter mean (p = 0.009), and mean diary score, including pain, itching, burning, and sleep disturbances (p = 0.016).

Discussion
Most of the presented studies demonstrated the adverse impact of smoking on OS, tumor control, QoL, treatment toxicity, and the incidence of SPC in cancer patients managed with RT. Detrimental effects of smoking concern definitive and postoperative (CH)RT in both smoking-related and unrelated malignancies. Early cessation after cancer diagnosis improves the efficacy of RT and QoL and decreases the risk of SPC and mortality from other tobacco-related diseases.
Somewhat surprisingly, as opposed to other malignancies, the detrimental effect of tobacco smoking on OS in NSCLC was inconsistent. This may be due to a very high prevalence of tobacco smoking in this group, making comparisons with much less numerous groups of non-smokers challenging. More uniform, both in lung cancer and other malignancies, was the adverse impact of continued smoking on the risk of SPC and treatment tolerance.
The harmful effect of smoking on treatment toxicity and induction of SPC is well recognized and applies to various malignancies and therapies. RT is a local treatment modality, and its beneficial effects are basically via locoregional tumor control. Many studies presented in this review indicate that smoking may adversely impact locoregional tumor control. The mechanisms underlying this effect remain uncertain. One of the proposed reasons is decreased tumor oxygenation, a critical factor for the efficacy of RT [67]. This effect may be caused by smoking-induced respiratory insufficiency before RT, exacerbated by increased carboxyhemoglobin levels in patients who continued smoking during RT [68].
We are aware of several limitations of our work. First, due to a large body and diversity of literature data, we have not attempted a comprehensive subject analysis. Nevertheless, we have done our best to objectively present studies that did and did not demonstrate a negative impact of tobacco smoking. Second, almost all presented studies are observational and retrospective, thus some potentially relevant data might have been missed. In the majority of studies, the self-reported smoking status was not validated biochemically. Some studies included mixed populations of patients who did and did not receive RT, and the subset analyses might have insufficient power to detect significant differences. In several studies the number of patients who quit or continued smoking during treatment, as well as quantitative tobacco exposure (PY), was missing. Many studies were too small to detect small differences between particular patient populations. Finally, the studies varied in duration of follow-up, analysis methods, and endpoints.

Conclusions
Tobacco smoking is probably the strongest modifiable factor affecting the outcome of cancer treatment. To better characterize its effect on RT outcomes, it is advisable to document a thorough smoking history in routine clinical practice and prospective clinical studies. Patients should be aware of smoking hazards and be advised when they and their families are being counseled. Healthcare professionals should use the cancer diagnosis and treatment as a teachable moment and recommend immediate smoking cessation [69]. Wherever possible, cancer patients should undergo an intensive smoking cessation program, including behavioral and pharmacologic therapy.