Role of Inflammation Gene Polymorphisms on Pain Severity in Lung Cancer Patients

Pain is one of the most devastating, persistent, and incapacitating symptoms in patients with lung cancer. Patients with advanced lung cancer suffer from significantly higher levels of physical and mental symptoms compared with patients with most other solid tumors. As many as 80% of patients with newly diagnosed lung cancer present with pain before any cancer treatment and of whom 17% report pain of severe intensity (1). Severe pain is reported by 41% of patients with advanced lung cancer (2). Cancer pain often occurs at multiple sites and duration can extend from months to years (3). Because of its high prevalence and the frequency with which patients with lung cancer present in an incurable stage, symptom management is a large component of the care of these patients.

Among cancer patients, chronic inflammation acts as a tumor promoter, resulting in aggressive tumor growth and spread. Many of the same inflammatory factors that promote tumor growth are also hypothesized to function as pain modulators not only in inflamed tissues but also in damaged peripheral nerves. The activation of inflammatory cells, for example, is classically associated with pain, heat, redness, swelling, and loss of function. It is now suggested that, following tissue damage or inflammation, inflammatory molecules including cytokines and chemokines also directly sensitize the peripheral terminals of sensory nerves (peripheral sensitization), thus lowering their pain activation threshold (4-6). Elevated cytokine levels, such as interleukin (IL)-6 and IL-8, are observed in patients with chronic pain conditions including back pain (7), post-herpetic neuralgia (8), and unstable angina. IL-1 and IL-2 levels have likewise been implicated in pain response (9, 10) and suggested to contribute to variation in postoperative morphine requirements (11) and in complex regional pain syndrome (12). IL-4 is correlated with the presence of chronic widespread pain (13) and the association of IL-10 level with pain and its potential role in pain therapy has also been suggested (14-17). Tumor necrosis factor-α (TNF-α) has an important role in cancer-related symptoms including pain facilitation and enhancement (18-20). The prolonged presence of increased levels of IFN-γ in the central nervous system contributes to the generation of central sensitization and persistent pain by reducing inhibitory tone in the dorsal horn (21-23). Taken together, these studies provide evidence of the critical role of the immune system in chronic pain states.

Single nucleotide polymorphisms (SNP) in the inflammation genes have been shown to alter their expressions or functions and thus may be associated with an altered risk for pain severity. Indeed, our group and others have shown polymorphisms in IL-6, TNF-α, and IL-8 to influence pain severity (1, 2). However, these studies only assessed one or a few select candidate genes at a time. Given that pain is a complex trait, multiple genes are likely to influence vulnerability to pain. Therefore, a pathway-based genotyping approach, which assesses polymorphisms in several genes that interact in the same pathway, may provide more robust results. In this study, we therefore evaluated a comprehensive panel of 59 SNPs in 37 inflammation genes in newly diagnosed non-Hispanic Caucasian lung cancer patients (n = 667) and assessed their association with pain severity. We also assessed the extent to which clinical and demographic factors explain pain severity in this population. Because genetic polymorphisms are stable markers, understanding the extent to which genetic variability plays a role in cancer-related pain may prove useful in identifying patients at high risk for pain and, importantly, could help in understanding patients who might benefit most from symptom intervention and ultimately in developing personalized and more effective pain therapies.

The study sample was drawn from an ongoing previously described case-control study of lung cancer (24). Case patients with newly diagnosed histologically confirmed non–small cell lung cancer were recruited at the time of initial registration at M. D. Anderson Cancer Center before initiation of any cancer treatment. There were no restrictions with regard to age, sex, ethnicity, or disease stage. All cases were residents of the United States. The overall response rate for the study was 80%. For this analysis, we used data from patient enrolled from 1999 to 2005 and for whom pain and genetic data were available. Because of issues associated with population stratification, we focused our analyses on 667 White Caucasian patients. This study was approved by the institutional review board at M. D. Anderson Cancer Center and all participants provided written informed consent.

Trained M. D. Anderson Cancer Center staff interviewers collected demographic, clinical, and symptom data before initiation of radiotherapy or chemotherapy. Patients rated their pain on an 11-point numeric scale (0 = “no pain” and 10 = “pain as bad as you can imagine”; ref. 25), a standardized method for assessing pain. Because studies show a high correlation between depression, fatigue, and pain, we also assessed depressed mood and fatigue using the following items: “During the past 4 weeks, have you felt downhearted and blue?” and “During the past 4 weeks, did you have a lot of energy?” These items were taken from the SF-12. The SF-12 is a validated measure of quality of life and is extensively used in studies of cancer patients (26-29). Data including stage of disease and history of comorbid conditions (heart disease, stroke, diabetes, etc.) were abstracted from patients' charts.

Charts were reviewed for information on opioid dose by a supportive care specialist (S.Y.). Due to the different types of opioids reported, we translated the daily opioid dose to a standardized measure, morphine equivalent daily dose (MEDD). We used the conversion factors shown in Table 1 to calculate the total dose of opioids.

After the interview was completed, a 40 mL blood sample was drawn into coded heparinized tubes. Genomic DNA was extracted from peripheral blood lymphocytes by proteinase K digestion followed by isopropanol extraction and ethanol precipitation. DNA samples were stored at −80°C. We selected for genotyping SNPs in immune response genes that met at least two of three criteria: (a) minor allele frequency of at least 5%; (b) location in the promoter, untranslated region, or coding region of the gene; and (c) previous report of an association with pain severity. All SNPs were genotyped using SNPlex, a technology developed by Applied Biosystems that enables simultaneous genotyping of up to 48 SNPs in a single tube using an oligonucleotide ligation assay and described previously (30).

Descriptive statistics were used to summarize patient characteristics. The Kolmogorov-Smirnov Z test was used to assess normality distribution for pain severity. Because normality was not met, we used the National Comprehensive Cancer Network cutoff score for severe pain (31). (A score ≥7 is considered as a pain emergency and treatment is initiated with short-acting opioids.)

We used logistic regression to assess associations between severe pain status and demographic (age and sex), clinical (stage of disease), and symptom (depressed mood and fatigue) variables. Variables found to be associated with severe pain at P < 0.05 were included in subsequent analyses.

We used multivariable logistic regression to assess associations between severe pain status and each SNP, adjusting for demographic (age and sex), clinical (stage of disease), and symptom (depressed mood and fatigue) variables found to be associated with severe pain. We focused on SNPs for which there was a statistically significant (P < 0.05) effect in an additive model (trend in pain risk with increasing copies of the less common, “mutant” allele) or for which there was also a significant association with severe pain for the mutant allele under a dominant model. We also examined recessive models (Appendix A).

Because there is a high degree of linkage disequilibrium between the three PTGS2 SNPs (D' was 0.99 between exon10-90C>T and exon10+837T>C, 0.95 between exon10+837T>C and -765G>C, and 0.64 between exon10-90C>T and -765G>C), we inferred the haplotypes consisting of these three SNPs for each patient using the available software PHASE version 2.1.1 (32). We assessed for significant associations using the two-sided binomial exact test.

There were a total of 677 White Caucasian patients with previously untreated and histologically confirmed non–small cell lung cancer. Mean (SD) age was 61 (12) years. There was about an equal distribution of the sample between early stage (stage I-IIIA; n = 325) and late stage (stage IIIB-IV; n = 321) of disease. There were more men (n = 351) than women (n = 326) and hypertension was the most prevalent comorbid condition.

Sixteen percent of the patients reported severe pain. MEDD (Table 2) computed as the total dose of opioids from the analgesic use reported at the time of presentation was between 0 and 1,000 mg/24 h; mean (SD) of 6.05 (47.25). As expected, Table 2 shows that severe pain was more prevalent among those with advanced stage of disease (OR, 2.34; 95% CI, 1.50-3.65; P = 0.001), younger age (OR, 2.10; 95% CI, 1.32-3.30; P = 0.002), reports of depressed mood (OR, 3.68; 95% CI, 1.96-6.93; P = 0.001), and fatigue (OR, 3.72; 95% CI, 2.36-5.87; P = 0.001). There was a borderline association for sex (females: OR, 1.43; 95% CI, 0.99-2.16; P = 0.06).

We evaluated 59 SNPs in the 37 immune response genes, adjusting for the nongenetic correlates (stage of disease, age, sex, MEDD, fatigue, and depressed mood). We observed that patients with CC genotypes for PTGS2 exon10+837T>C (rs5275) were at lower risk for severe pain (OR, 0.33; 95% CI, 0.11-0.97) and an additive model for TNFα −308GA (rs1800629; OR, 1.67; 95% CI, 1.08-2.58) and NFKBIA Ex6+50C>T (rs8904) was predictive of severe pain (OR, 0.64; 95% CI, 0.43-0.93). In the multigene model for severe pain (Table 3B), we found that only TNFα −308GA significantly predicted severe pain. PTGS2 exon10+837T>C and NFKBIA Ex6+50C>T were borderline significant (P < 0.06).

We also assessed the extent to which the number of protective alleles influences pain severity by combining the allelic information for TNFα −308GA, PTGS2 exon10+837T>C, and NFKBIA Ex6+50C>T (Table 4A). Table 4 shows that each protective allele decreased the risk for severe pain by as much as 38% (Table 4B) even after adjustment for the nongenetic variables (MEDD, stage of disease, age, sex, depressed mood, and fatigue), suggesting a gene-dose effect.

When we assessed for association between a particular haplotype/diplotype status and pain severity, we did not find significant association between the PTGS2 haplotypes and severe pain (Table 5).

To address the multiple comparison problem, we calculated the false-positive report probability (FPRP) for the SNPs PTGS2 exon10+837T>C (rs5275), TNFα −308GA (rs1800629), and NFKBIA Ex6+50C>T (rs8904) which were found to be significant. The FPRP is the probability that the significant finding is false (33). FPRP calculations depend on the observed P value for the association, prior probability that the association between the genetic variant and the disease is real, and the statistical power of the test. For our analyses, we assumed a range of prior probabilities from 0.01 to 0.10. For the statistical power calculations, we used observed ORs of each specific SNPs. The noteworthiness of an association is defined as having FPRP value below 0.5 for initial exploratory studies (33). Table 6 gives noteworthiness of our significant association under different prior probabilities. From this table, we see that our observed association is noteworthy for prior probabilities that are >0.05 (for TNFα −308GA and NFKBIA Ex6+50C>T) and 0.10 (for PTGS2 exon10+837T>C; considered to be moderate prior probability).

Although previous studies have shown the influence of inflammation-related genes on pain severity in several disease conditions, these studies only assessed a few candidate genes and with small sample sizes. In this study, we conducted a systematic assessment of the influence of a larger number of polymorphisms in inflammation-related genes on pain severity in a large sample of newly diagnosed, previously untreated patients with non–small cell lung cancer. We found that functional variants of the PTGS2 exon10+837T>C (rs5275), TNFα −308GA (rs1800629) and NFKBIA Ex6+50C>T (rs8904) contribute to pain severity. The most significant finding was that, in analysis of the joint effects, the number of observed protective genotypes was associated with a reduced risk in a dose-responsive manner, with each protective genotype reducing the risk for severe pain by as much as 38%.

We also observed a significant association with polymorphisms in TNFα −308GA and pain severity. The -308 polymorphism is a G→A substitution and reportedly affects gene expression, the rare A allele resulting in higher TNF production (34). TNF-α has been suggested to be critical for the development of inflammatory pain behavior in animal models. The novel therapeutic potential of TNF inhibitors has also been suggested for conditions such as brain cancer, epilepsy, and chronic pain (35-38). Anti-TNF therapy has also been shown to be profoundly analgesic, with an efficacy similar to that of cyclooxygenase-2 (COX-2) inhibition, and reduced astrocyte activity in collagen-induced arthritis (37).

Importantly, carriers of the homozygous variant genotype (CC) of PTGS2 exon10+837T>C exhibited significantly protective effect (OR, 0.32) for severe pain. Specifically, carriers of CC genotypes had 64% reduced risk for severe pain relative to carriers of the TT and TC genotypes even when demographic, clinical, and other symptom variables were taken into account.

The PTGS2 gene encodes the proinflammatory COX-2 enzyme. Exon10+837T>C of the PTGS2 gene is a functional SNP that modulates expression of COX-2. Subjects with the variant genotypes of exon10+837T>C were observed to have lower steady-state PTGS2 mRNA level than those with the homozygous wild-type (mean ± SE, 15.96 ± 2.82 versus 33.02 ± 14.66; ref. 39). COX-2 is inducible and up-regulated during an inflammatory response. COX-2 is rapidly induced by growth factors, cytokines, and proinflammatory molecules, and is involved in prostanoid production under acute and chronic inflammatory conditions as well as in neurodegenerative processes, ischemia, normal neuronal functioning, neurotoxicity, and synaptic plasticity (40). Peripheral elevation of COX-2 after tissue injury contributes to increased prostaglandin E (2) at the site of injury and leads to pain onset. Indeed, COX-2 is a therapeutic target for pain. Inhibition of COX-2 enzymatic activities is responsible for the anti-inflammatory properties of aspirin, indomethacin, and ibuprofen and related nonsteriodal anti-inflammatory drugs, such as Vioxx (rofecoxib) and Celebrex (celecoxib).

We found that polymorphisms in NFKBIA gene were also predictive of severe pain. NF-κB is activated on noxious stimulation and contributes to pain hypersensitivity by increasing the transcription of “pain-related” genes such as COX-2 and proinflammatory cytokines. Animal studies show that NF-κB inhibition attenuates the nociceptive response in models of neuropathic pain (41, 42). Intrathecal pretreatment of rats with NF-κB inhibitors reduced spinal NF-κB activation and subsequent expression of COX-2 mRNA, thereby suppressing hyperalgesia following unilateral hind paw inflammation (43).

Consistent with our previous studies (44), we found that depressed mood and fatigue were also significant correlates of pain. Several studies have addressed the relationship between depression, fatigue, and pain and found these symptoms to co-occur. Although the causal relationship between these symptoms remains debatable, studies have shown that symptoms, such as pain, are in fact associated with depressive disorders or psychologic distress and anxiety (45-47). It has also been hypothesized that a shared biological mechanism may underlie the co-occurrence of these symptoms (48). Among the implications of these findings is the need to address symptoms such as depressed mood and fatigue to improve on pain severity as well as study of potential common underlying genetic mechanisms for both pain and depressed mood.

Although this study has a relatively large number of patients, there remains concern about the issue of false-negative findings (failed to detect SNPs with small contribution to pain severity). One could also argue that pain is a heterogeneous outcome with a variety of causes; for example, neuropathic pain is different from pain related to pressure from a large tumor, which is different from pain related to stretching of a capsule; thus, a 0-10 pain severity/intensity rating is a global measure of pain, which does not delineate if the pain measured is of a neuropathic or nociceptive type of pain. However, evidence suggests that cancer pain is typically of a mixed pain mechanism, with only a small proportion of cancer patients suffering from pure neuropathic pain at diagnosis. A review of pain studies (49) in lung cancer patients, for example, found that neuropathic pain accounted for 30% (range, 25-32%) of cases, with nociceptive pain as the major pathophysiologic subtype in lung cancer pain. Furthermore, although neuropathic pain may occur due to a malignant invasion of neurologic structures (including pancoast tumors), neuropathic pain in cancer patients occurs as a late effect of treatment with Vinca alkaloids, taxanes, platinum-derived compounds, radiotherapy, or surgery. Given that our study focused on newly diagnosed lung cancer patients, who have not had any cancer treatment, misclassification of the type of cancer pain (of whether nociceptive or neuropathic) was greatly attenuated. We also acknowledge that there is more genetic variation for each gene than is captured in this study. The selective choice of SNPs for each gene limited our ability to perform more extensive haplotype analyses. In conducting the FPRP analyses, we found that our observed association is noteworthy for initial studies and therefore should be assessed in confirmatory studies.

In conclusion, despite advances in pain treatment and management for cancer, a significant number of patients continue to suffer from severe and persistent pain. Although epidemiologic, clinical, and psychologic factors have been shown to influence pain and its treatment, we have also shown in a preliminary fashion that variation in pain severity and pain treatment response may be partially attributed to host genetic variability. Future studies with larger cohorts are needed to validate our findings.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.