Opioids in pain management of mesothelioma and lung cancer patients

Abstract

Background. Lung cancer and mesothelioma are malignant tumors with generally dismal prognosis and therefore palliative pain treatment constitutes a challenge for the clinician. Objectives. The aim of this study was to compare the outcomes of pain treatment with opioids among mesothelioma and lung cancer patients treated for palliation and assess factors which confound to optimal treatment. Patients and methods. A sub-cohort of 373 lung cancer and 22 mesothelioma patients was identified in multi-center European Pharmacogenetic Opioid Study (EPOS) cohort. A nested case-control (1:4) setting was designed to estimate the pain and other covariates distinguishing 22 mesothelioma- (= cases) and 88 lung cancer patients (controls), analyzed using univariate- and multivariate conditional (fixed-effects) logistic regression models. Results. The mean total daily dose of opioids varied from 30.0 to 960.0 mg (mean 275, median 160 mg, SD 293) in mesothelioma, and from 10 to 5072 mg (mean 414, median 175, SD 788) in lung cancer patients (p = 0.420). In both groups, pain was mostly experienced as moderate and severe and it was frequently accompanied by depression, poor sleep, anxiety and fatigue. Four mesothelioma patients (18%) and seven lung cancer patients (10%) experienced complete pain relief with opioids by self-assessment. Assessments of pain severity by the patients and their physicians deviated significantly in mesothelioma (p = 0.039 McNemar test), as well as in lung cancer (p = 0.0001). In conditional logistic regression, no significant differences were found in distribution of pain covariates between lung cancer and mesothelioma patients. Conclusion. Pain perception by the patients was associated frequently with other symptoms and complete pain control with opioids was achieved only with minority of patients both with mesothelioma and advanced lung cancer. Adequate pain control requires continuous monitoring and tailoring the dose to patient's individual needs and tolerance, recognition of accompanying symptoms such as depression and poor sleep, and their management.

More than 10 000 annual new mesothelioma cases are reported worldwide. In Finland, there were 79 cases among males and 21 among females in 2007, implicating a clear (3:1) male preponderance [1]. Its incidence has been increasing during the past decades. Approximately 80% of mesothelioma cases can be directly linked to patient's occupation, i.e. exposure to asbestos, with latency periods of 20–50 years [2]. Lung cancer is among the most common cancers worldwide and remains to be the leading cause of cancer mortality [3].

Although mesothelioma represents a rare malignancy, it can be a clinical challenge due to treatment resistance and dismal prognosis. There is no effective cure except in rare cases amenable to radical surgery and multimodal treatment [4–6]. The one-year survival rate levels off at 40% after different treatment approaches, and the five-year survival rate is about 10%.

The five-year survival of lung cancer patients is slightly better but remains under 15%. Also for lung cancer patients with advanced disease, effective pain management constitutes an essential part of care. According to Potter and Higginson, the overall weighted mean prevalence of pain experienced by lung cancer patients was 47%, being even higher among the patients attending cancer treatment centers (65%) and in patients referred to palliative care services (76%) [7]. In another study on patients with different cancers, lung cancer patients had the worst burden of symptoms [8].

Recently, 2294 patients from 11 European countries participated in the European Pharmacogenetic Opioid Study (EPOS) [9]. In this cohort, both self-reports and observer assessments of pain and covariates are available from 1933 patients [10]. Altogether, 22 mesothelioma patients were identified, and for the present analysis, a nested case- control (1:4) setting was designed, including 88 age-and sex-matched lung cancer patients as controls. In this study the focus was on secondary outcomes as the original study had another aim. In addition to comparing the outcomes of pain control and management with opioids in mesothelioma and lung cancer patients under palliative care, the covariates (predictors) of pain were analyzed using univariate and multivariate conditional (fixed-effects) logistic regression models.

Patients and methods

Patients

The patients represent a sub-cohort of the EPOS study, in which 2294 cancer patients from 11 European countries were enrolled, as recently detailed [9]. In this cohort, 22 mesothelioma patients and 374 lung cancer patients were identified, comprising the subjects of the present study. Patients were recruited from 17 centers. The Pain and Palliation Research group, Norwegian University of Science and Technology organized the collection of the data. The assessment of symptoms and pain experience were reported both: 1) by the physician using an observer rating score of symptom intensities (none, mild, moderate or severe) during the past 24 hours [11]; and 2) by the patient filling and EORTC-QoL C30 questionnaires. For symptoms assessment of fatigue, nausea, depression, dyspnoe, anxiety, constipation and sleeplessness were defined as reported by the patient as severe or moderate. All use of opioids and other medications were recorded in detail. Opioid doses were calculated into equipotent daily oral morphine doses [9]. Performance status was assessed using the Karnofsky criteria [12]. In the present study, pain and its covariates in mesothelioma patients and lung cancer patients were compared in a matched case-control setting, nested within the EPOS sub-cohort, as described in the following.

Statistical analysis

Statistical analyses were done using two statistical software packages: SPSS 19.0.1. for Windows (IBM, NY, USA) and STATA/SE 11.2 software (STATA Corp., Texas, USA). Frequency tables were analyzed using the χ²-test, with Fischer’s exact test to assess the significance levels between categorical variables. Differences in the means of continuous variables were analyzed using the non-parametric tests (Mann-Whitney, Kruskal-Wallis), depending on their normal distribution. Crude Odds Ratios (OR) and their 95% Confidence Intervals (CI) were calculated for predictors of pain, using univariate logistic regression. For each 22 mesothelioma patients, four ‘control’ patients (n = 88) were selected among the EPOS lung cancer patients (n = 374), matched strictly by age and gender, to create a nested case-control (1:4) setting. In this nested case-control setting, conditional logistic (fixed-effects) regression analysis was used to evaluate the association of pain covariates with mesothelioma (using lung cancer as the reference) in both univariate and multivariate models. Results are expressed as crude OR (95% CI) and adjusted OR, controlled for all other covariates in the model. All tests were two-sided and interpreted significant at p= 0.05 level.

Results

Clinical characteristics of the patients included in the present series are summarized in Table I. There was a clear male preponderance, 16:6 and 64:24 for mesothelioma and lung cancer patients, respectively. The mean age of patients was 63 (SD 9.3) years. The mean Karnofsky performance score for mesothelioma and lung cancer patients, respectively, was 60 (range 30–90) and 60 (range 20–90), indicating that most patients had poor/decreased performance. Most patients were also anemic (Hb < 12.0 g/dl). Thirteen mesothelioma patients were hospitalized during the data recording, whereas nine patients were observed in an outpatient clinic. The corresponding numbers for lung cancer patients were 78 and 10. Mean time since diagnosis was 13.4 months (range: 0–95) for mesothelioma, and 9.9 months (range: 0–106) for lung cancer patients (p = 0.367), with the mean time since initiation of opioid treatment being 8.8 months (range: 0–69) and 3.3 months (range: 0–48), respectively (p = 0.119, Mann-Whitney.). Seventeen mesothelioma patients and 60 lung cancer patients were treated with oral opioids (morphine, oxycodone, methadone and hydromorphone), while five and 20 patients, respectively, obtained the treatment with a trans-dermal patch (fentanyl). In addition, four lung cancer patients received subcutaneous and three patients received intravenous opioids.

Table I. Characteristics of the patients with mesothelioma and lung cancer.

	Mesothelioma N= 22	Lung cancer N= 88
Gender: Male Female	16 (73%) 6 (23%)	64 (73%) 24 (23%)
Age	63 years, SD= 9.6 (range 41–79)	63 years, SD = 9.6 (range 37–80)
Karnofsky (Mean)	63, SD = 16.6 (range 30–90)	56, SD= 16.2 (range 20–90)
Hb level (Mean) < 120 g/l	11.6, SD= 1.6 13 (59%)	11.6, SD= 1.6 52 (59%)
Mean time (range) since diagnosis months	13.4 (range 0–95)	9.9 (range 0–106)
Mean time (range) since initiation of opioid treatments months	8.8 (0–69)	3.3 (0–48)

Table II shows pain characteristics of the patients in the two groups. In most cases, pain was moderate and severe in intensity. Concordance (agreement) in pain assessment between the mesothelioma patients and providers was only fair (Kappa= 0.203, p = 0.039). In lung cancer patients, these two assessments were even more discordant (Kappa= 0.188, p= 0.018, poor agreement). In pair-wise comparison, assessments of pain severity by the patient and the physician deviated significantly in both mesothelioma (p = 0.039 McNemar test) (12/22 concordant pairs), and lung cancer (p = 0.0001) (31/70 concordant pairs).

Table II. Key characteristics of pain in mesothelioma and lung cancer patients.

	Mesothelioma N = 22	%	Lung cancer N= 88	%	Significance@
Pain category Bone/soft-tissue pain Mixed pain Visceral pain Neuropathic pain	8 11 2 1	36 50 9 5	41 36 7 2	47 41 8 2	p = 0.765
Localization of pain (main site) Thorax front/abdomen Abdominal/pelvic	12 10	55 46	38 26	59 36	p = 0.453
Patient assessment/Have you had pain (EORTC QoL)* Not at all A little Quite a bit Very much	1 6 11 3	5 27 50 14	5 18 27 19	7 26 39 28	p= 0.619
Physician's assessment of pain** Not at all A little Quite a bit Very much	1 12 7 2	5 55 32 9	13 38 26 10	15 44 30 11	p= 0.602

*Assessed by 69/88 patients, **Assessed for 87 patients; @Fisher's exact test.

The majority (n = 50) of the lung cancer patients (57%) received also systemic steroid medication mostly (n = 32, 36%) with dexamethasone, and another 19 patients (22%) needed additional paracetamol. For 52 (59%) patients reporting moderate to severe constipation, laxatives were given. Altogether, 24 (27%) patients received treatment with antidepressants, and nine (10%) patients with hypnotics. In the mesothelioma group, 11 patients (50%) received systemic steroid medication, mostly (n = 9, 41%) with dexamethasone, while four patients (18%) were treated with additional paracetamol. For 11 patients reporting moderate to severe constipation, laxatives were given, six patients received antidepressants and five obtained hypnotics as well.

Table III lists the covariates associated with moderate to severe pain (outcome measure) among lung cancer patients in univariate regression analysis. Younger age (OR = 2.7, 95% CI 1.1–6.7), poor Karnofsky (OR = 3.50, 95% CI 1.2–9.9), sleeplessness (OR = 3.40, 95% CI 1.4–8.4), as well as persistence of breakthrough pain (OR = 2.92, 95% CI 1.2–7.3) were the significant covariates.

Table III. Determinants of pain in univariate analysis of lung cancer patients. Number (%).

Variables	Moderate to severe pain		Crude OR	95% CI	Significance p
	Present	Absent
Gender: Male Female	23 (36) 13 (54)	40 (64) 11 (46)	0.49	0.2–1.3	0.137
Age: Below median Above median	26 (51) 10 (28)	25 (49) 26 (72)	2.70	1.1–6.7	0.029
Karnofsky: Below median Above median	30 (50) 6 (22)	30 (50) 21 (78)	3.50	1.2–9.9	0.013
Hb: Below 120 Above 120	21 (41) 15 (42)	30 (59) 21 (58)	0.98	0.4–2.3	0.964
Fatigue: Yes No	33 (45) 3 (23)	41 (55) 10 (77)	2.68	0.7–10.6	0.134
Nausea: No Yes	12 (52) 24 (38)	11 (48) 40 (62)	1.81	0.7–4.8	0.223
Depression: Yes No	20 (39) 16 (46)	31 (61) 19 (54)	0.77	0.3–1.8	0.549
Dyspnoea: Yes No	19 (40) 17 (44)	29 (60) 22 (56)	0.85	0.4–1.9	0.706
Anxiety: Yes No	28 (46) 8 (31)	33 (54) 18 (69)	1.90	0.7–5.1	0.185
Constipation: Yes No	19 (45) 16 (36)	23 (55) 28 (64)	1.45	0.6–3.4	0.402
Sleeplessness: Yes No	23 (56) 12 (27)	18 (44) 32 (73)	3.40	1.4–8.4	0.007
Breakthrough pain: Yes No	26 (52) 10 (27)	24 (48) 27 (73)	2.92	1.2–7.3	0.018
Anorexia: Yes No	17 (40) 19 (43)	26 (60) 25 (57)	0.86	0.4–2.0	0.730

*χ²-test, Likelihood ratio statistic (two-sided).

Similar analysis was made for the mesothelioma patients (Table IV). The above listed four covariates seemed to emerge as most apparent ones, but the limited number of patients precluded any of them to reach statistical significance.

Table IV. Determinants of pain in univariate analysis of mesothelioma patients.

Variables	Moderate to severe pain		Crude OR	95% CI	Significance p
	Present	Absent
Gender: Male Female	6 (38) 3 (50)	10 (63) 3 (50)	0.60	0.1–3.9	0.655
Age: Below median Above median	6 (46) 3 (33)	7 (54) 7 (67)	1.71	0.3–9.9	0.674
Karnofsky: Below median Above median	6 (67) 3 (30)	3 (33) 7 (70)	2.33	0.4–13.7	0.415
Hb: Below 120 Above 120	5 (39) 4 (44)	8 (62) 5 (56)	0.78	0.1–4.4	1.000
Fatigue: Yes No	9 (43) 0 (0)	12 (57) 1 (100)	0.57	0.4–0.8	1.000
Nausea: No Yes	4 (67) 5 (31)	2 (33) 11 (69)	4.40	0.6–32.5	0.178
Depression: Yes No	6 (43) 3 (38)	8 (57) 5 (63)	1.25	0.2–7.4	1.000
Dyspnoea: Yes No	5 (36) 4 (50)	9 (64) 4 (50)	0.56	0.1–3.3	0. 662
Anxiety: Yes No	9 (50) 0 (0)	9 (50) 4 (100)	0.50	0.3–0.8	0.115
Constipation: Yes No	7 (50) 2 (25)	7 (50) 6 (75)	3.00	0.4–20.3	0.380
Sleeplessness: Yes No	5 (46) 4 (36)	6 (55) 7 (64)	1.46	0.3–8.1	1.000
Breakthrough pain: Yes No	7 (54) 2 (22)	6 (46) 7 (78)	4.1	0.6–27.7	0.203
Anorexia: Yes No	7 (58) 2 (20)	5 (42) 8 (80)	5.6	0.8–38.5	0.099

*Fisher's exact test (two-sided).

Table V summarizes the results of the conditional logistic regression analysis for the pain covariates associated with mesothelioma, using lung cancer patients as a reference. In univariate analysis, all covariates obtained OR > 1.00, implicating that despite their higher mean Karnofsky (OR = 1.88) and slightly higher proportion of patients with better hemoglobin levels (Hb > 12.0 cut-off, OR = 1.04), mesothelioma patients experienced all the tested pain covariates more frequently than did the lung cancer patients (= reference). This difference is most marked for reporting fatigue, crude OR = 3.44 (95% CI 0.5–24.4). When all tested covariates were entered in the multivariate model, some of them are confounded (i.e. adjusted OR lower than crude OR), while some others were negatively confounded (i.e. adjusted OR higher than crude OR). This is particularly the case with fatigue, reaching OR = 5.22 (95% CI 0.8–32.1) (p = 0.075).

Table V. Pain and other covariates associated with mesothelioma and lung cancer patients in a nested case-control setting*.

	Dependent variable: Mesothelioma (cases) or lung cancer (controls)
Covariates	Crude OR	95% CI	p	@Adjusted OR	95% CI	p
Age (at enrolment)	MV			MV
Gender	MV			MV
Karnowsky (median cut-off) (> median: reference)	1.91	0.6–6.0	p= 0.266	2.29	0.6–8.2	p= 0.203
Hb (120 cut-off) (> 120: reference)	1.00	0.4–2.7	p= 1.000	1.08	0.4–3.3	p= 0.883
Fatigue (yes: reference)	3.44	0.5–24.4	p= 0.215	5.16	0.8–31.3	p= 0.074
Nausea (yes: reference)	1.05	0.4–3.2	p= 0.920	0.70	0.2–2.4	p= 0.574
Depression (yes: reference)	1.17	0.5–2.9	p= 0.729	0.90	0.3–2.9	p= 0.869
Dyspnoea (yes: reference)	1.37	0.5–3.9	p= 0.540	1.82	0.6–5.7	p = 0.300
Anxiety (yes: reference)	1.81	0.5–6.1	p= 0.335	1.46	0.4–5.6	p= 0.581
Constipation (yes: reference)	1.72	0.6–4.8	p= 0.295	1.71	0.6–5.2	p= 0.335
Sleeplessness (yes: reference)	1.11	0.4–2.8	p= 0.822	1.28	0.5–3.3	p= 0.605
Breakthrough pain (yes: reference)	1.05	0.4–3.1	p= 0.927	1.10	0.3–3.7	p= 0.869
Anorexia (yes: reference)	1.19	0.4–3.3	p= 0.730	0.82	0.2–3.0	p= 0.768
Total opioid dose (median cut-off) (> median: reference)	1.00	0.4–2.4	p= 1.000	1.04	0.4–2.7	p= 0.931

*Analyzed by conditional logistic regression models, robust CVE; MV, matching variable; NC, not computable; @adjusted for all covariates in the model except for the matching variables.

Discussion

Lung cancer and mesothelioma remain to be therapeutic challenges. The majority of patients die of advancing cancer, mostly with poor general condition and multiple symptoms. Optimal pain management is among the key goals in palliative care. Our results show that complete pain control with opioids was achieved in less than 20% of these patients. Therefore the assessment of outcome in pain management needs to be further developed, including comprehensive approach taking into account other symptoms which are frequently associated, such as depression, anxiety and poor sleep. It is essential that physician's assessment on symptoms is concordant with the patient's perception of pain and symptoms related to total suffering. Similar issues in lack of adequacy of pain management were observed both in mesothelioma and lung cancer. Monitoring of symptoms need to base on patients’ reporting and alternative modes of administration should be considered as appropriate.

When mesothelioma originates from the pleural cavity, the first symptom may be a troublesome discomfort or pain in the chest area or in the back. If the tumor originates in the peritoneal lining of abdominal cavity, the first symptom is abdominal or pelvic discomfort [13]. Along with disease progression and destruction of soft tissue and nearby nerves, the patient experiences increasing discomfort and pain. General pain was reported by one third of these patients and chest pain by 25% of the patients in a large clinical study [14]. Mesothelioma has one of the most dismal prognoses among cancers due to often advanced stage at diagnosis, lack of specific symptoms or signs to enable screening and early detection, and high level of resistance to current treatment modalities [15]. For optimal pain control, there may be need for invasive methods such as epidural or spinal administration of analgesics may have to be considered due to infiltrating spread of this tumor.

In lung cancer, the overall prognosis is poor except in early cases, which can be cured by surgery or radiotherapy. Thus, medical symptom management including opioids plays a major role and actually forms the backbone of adequate palliative treatment of patients suffering from pain in advanced mesothelioma or lung cancer. In a cross-sectional study of consecutive patients undergoing radiation treatment for head/neck, lung or prostate cancers, the intensity of most severe pain was significantly greater in lung cancer patients [16]. More than one half of all patients experienced high levels of fatigue and pain. Fatigue and frequency of pain continued to be the most distressing symptoms at all times among patients receiving treatment for lung cancer as a whole [17]. Multisymptom targeting tailored according to patient response might be the way to obtain optimal symptom control and to interrupt the ‘circulus vitiosus’ of multiple symptoms.

The time since diagnosis was quite similar in both patient groups. However, the initiation of opioids was more recent among lung cancer patients and this may influence the outcome. In the present series pain among mesothelioma patients was mostly of mixed character, while neuropathic pain was less common. This, together with shorter time to opioid treatment than among lung cancer patients may explain the somewhat better response to opioid treatment achieved with moderate opioid dose and parenteral (or cutaneous) administration compared to response in lung cancer. Pain in lung cancer was mostly of bone, soft tissue and mixed character, where also palliative radiotherapy can provide good symptom control.

In cancer, pain is not a solitary symptom per se, but it is often associated with other symptoms such as fatigue, poor sleep and depression [18], significantly affecting the QoL of these patients. In the present series, patients tended to have multiple syndromes even when pain was limited to thoracic or upper abdominal area. Anemia was common. Pain, poor sleep, fatigue and depression are interlinked and decrease the QoL. An extensive set of these pain-related symptoms (pain covariates) were recorded in the present study, and the two diseases were compared in a matched case-control design. In univariate analysis (run separately for mesothelioma and lung cancer) (Tables III and IV), four of these covariates were significantly associated with moderate-to-severe pain in lung cancer patients: younger age, poor Karnofsky, sleeplessness and persistence of breakthrough pain. These same covariates also appeared as candidates of significant predictors of pain in mesothelioma patients.

Underestimation of symptom intensity by health care providers is not rare and it can significantly increase the risk of inadequate treatment [19], as also confirmed in the present study. Laugsand and co-authors described the contributing factors, which are related both to the patient characteristics and the provider's background [19]. In the present study, as compared with the patient's individual perception, the physician frequently underestimated the pain severity, and only rarely were the two assessments concordant (Table II). This applied equally well to mesothelioma and lung cancer, which is not unexpected, because these two diseases did not significantly deviate from each other in their profile of pain covariates as shown with the nested case-control setting. The practical implications of these results necessarily include further development and refinement of the assessment tools by which the patients can more accurately than at present communicate their individual perceptions of pain and related symptoms, which add to the perception of pain [18]. This is the prerequisite for optimal response to the patient's needs in adequate pain management, which needs to be individually tailored, as also indicated by the wide dose range of opioids required by these patients in the present study.

Despite its main limitations, which are cross-sectional character, the small number of mesothelioma patients with limited power of conclusions and different time to opioids treatment between the groups, this study gives valuable data on the potential problems encountered in pain treatment with opioids both in mesothelioma and lung cancer. In general, the risk of practically all these variables to associate with mesothelioma was higher than with lung cancer (= reference), i.e. OR> 1.0, implicating that these pain-associated symptoms were more likely to be present in mesothelioma patients as compared with lung cancer patients. Fatigue is well known to associate with advanced cancer, but it may also increase due to suboptimal pain treatment.

The aim of palliative pain treatment is to sustain the patient's ability to maintain reasonable QoL, which makes it possible to live as independently as possible and preferably in home care. Monitoring the effects and carefully tailoring opioid dose to meet the individual patient's requirements and tolerance can result in more adequate pain control than seen in this study. Opioid treatment needs to be tailored on individual basis considering also other than oral administration modes, confounding symptoms need to be treated and the adverse effects should be treated as needed or even before they occur. In the present study, a few patients received cortisone and paracetamol, which can contribute to the success in pain management. However, laxatives were not routinely administered, although this is one of the basic recommendations for patients receiving opioid treatment.

In conclusion, pain among patients with both mesothelioma and advanced lung cancer was not optimally controlled since only minority of patients in both groups achieved complete pain relief.

The present study highlights the fact that patient's own assessment and other symptoms related to pain should be given more attention, and pain treatment should be tailored along with the treatment of confounding symptoms on an individual basis. In addition to management of pain, the psychosocial and spiritual support [20] should be an essential part of comprehensive palliative management to overcome the distress and despair associated with advanced mesothelioma and advanced lung cancer.

Acknowledgements

The authors are grateful to all the researchers involved in The European Pharmacogenetic Opioid Study (EPOS).

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

This study was supported by grants from the European Palliative Care Research Collaborative (EPCRC), the European Commission's Sixth Framework Program, the Norwegian University of Science and Technology and the Research Council of Norway and South Western Finland Hospital District (E.S.).