Morphine and its congener opioids are the main therapy for severe pain in cancer. However, chronic morphine treatment stimulates angiogenesis and tumour growth in mice. We examined if celecoxib (a cyclooxygenase-2 (COX-2) inhibitor) prevents morphine-induced tumour growth without compromising analgesia. The effect of chronic treatment with celecoxib (by gavage) and/or morphine (subcutaneously), or PBS on tumour prostaglandin E2 (PGE2), COX-2, angiogenesis, tumour growth, metastasis, pain behaviour and survival was determined in a highly invasive SCK breast cancer model in A/J mice. Two weeks of chronic morphine treatment at clinically relevant doses stimulates COX-2 and PGE2 (4.5-fold compared to vehicle alone) and angiogenesis in breast tumours in mice. This is accompanied by increased tumour weight (∼35%) and increased metastasis and reduced survival. Co-administration of celecoxib prevents these morphine-induced effects. In addition, morphine and celecoxib together provided better analgesia than either agent alone. Celecoxib prevents morphine-induced stimulation of COX-2, PGE2, angiogenesis, tumour growth, metastasis and mortality without compromising analgesia in a murine breast cancer model. In fact, the combination provided significantly better analgesia than with morphine or celecoxib alone. Clinical trials of this combination for analgesia in chronic and severe pain in cancer are warranted.
Opioids are the mainstay of treatment for severe pain in advanced cancer, including metastatic breast cancer (Levy, 1996; Mantyh, 2006). Although opioids act directly on the central nervous system to relieve pain, their activities on peripheral tissues are responsible for many of the secondary complications associated with the management of cancer pain. A recently recognised peripheral effect of opioids and their receptors is the promotion of angiogenesis-dependent tumour growth. Opioids at physiologically relevant concentrations promote angiogenesis in vitro, and in breast cancer and wound healing in rodents (Gupta et al, 2002; Poonawala et al, 2005; Singleton et al, 2006). Although naloxone and naltrexone can inhibit tumour growth in rodents (Zagon and McLaughlin, 1983a; Koo et al, 1996; Gupta et al, 2002), opioid receptor antagonists cannot be used to counteract the unwanted effects of opioids without also compromising analgesia in the clinical setting. Therefore, it is important to identify agents that spare the analgesic effect of opioids while inhibiting their angiogenic effect.
One key mechanism of morphine-induced angiogenesis is nitric oxide (NO)-dependent MAPK phosphorylation and endothelial proliferation (Gupta et al, 2002; Poonawala et al, 2005). Morphine stimulates NO in endothelium and chronic morphine treatment increases NO synthase (NOS), NO and cyclooxygenase-2 (COX-2) in mouse kidney (Stefano et al, 1995; Arerangaiah et al, 2007). Nitric oxide stimulates the enzymatic activity of COX (Salvemini et al, 1993), and activated COX in turn increases prostaglandin E2 (PGE2) production (Salvemini et al, 1993, 1994; Nedelec et al, 2001; Birnbaum et al, 2005). Prostaglandin E2 promotes angiogenesis and tumour progression, and blockade of COX-2 activity inhibits angiogenesis and tumour progression (Griffin et al, 2002; Leahy et al, 2002; Chang et al, 2004). On the other hand, prostaglandins are also implicated in processing pain (Malmberg and Yaksh, 1992; Julius and Basbaum, 2001; Samad et al, 2001). Indeed, COX-2 inhibitors inhibit the development of morphine tolerance (Wong et al, 2000).
It is likely that the products of COX-2 such as PGE2 produced in tumour cells act on the tumour endothelium and promote angiogenesis. Therefore, morphine-induced tumour growth may in part be due to the upregulation of COX-2. We hypothesised that the prolonged use of opioids in cancer stimulates COX-2 and PGE2, leading to angiogenesis-dependent tumour progression and metastasis and mortality. This raises the possibility that highly selective COX-2 inhibitors such as celecoxib may block morphine-induced angiogenesis and tumour growth without compromising analgesia. We examined the effects of celecoxib on opioid-induced COX-2 expression, pain, angiogenesis, tumour progression, metastasis and survival in a murine breast cancer model. We demonstrate that celecoxib prevents opioid-induced tumour progression and metastasis and promotes survival, apparently without impairing the analgesic effect of opioids.
Materials and methods
Eight- to twelve-week-old A/J mice (The Jackson Laboratory, Ann Harbor, ME, USA) were maintained under controlled environmental conditions (12-h light to dark cycle, at 23°C) in our AAALAC-accredited facility and provided with standard laboratory food and water ad libitum. All experiments were performed according to the protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Minnesota. Mice were euthanised using compressed CO2 from the gas cylinders at the conclusion of the duration of the treatment and/or if they appeared moribund (unable to reach food and water and impaired mobility).
Drugs and chemicals
Pharmacological-grade morphine (Baxter Esilerderle Mfd. Healthcare Corporation, Cherry Hill, NJ, USA), and Celecoxib (celecoxib capsules, Pfizer Inc., New York, NY, USA) were used. All other chemicals were obtained from Sigma Chemicals (St Louis, MO, USA), and cell culture reagents were obtained from Invitrogen unless otherwise mentioned.
SCK mammary carcinoma was generated in rodents using breast cancer cells derived from a tumour that spontaneously arose in a female A/J mouse (Jackson Laboratories) in our laboratory (Griffin et al, 2002). These tumours grow to ∼1000 mm3 in 12 days, and they are highly vascular, invasive and metastasise to the lungs, resulting in the death of the animals starting at about 15 days after tumour-cell injection. SCK tumour cells (2 × 105) in 0.05 ml of serum-free medium were injected subcutaneously into the right hind thigh of A/J mice. Palpable and measurable tumours appeared 4–5 days after injection.
Tumour-bearing mice were randomly assigned to experimental groups 1 day after tumour cell injection. Treatments for each experiment were administered twice daily.
To examine the effect of celecoxib on tumour PGE2, angiogenesis, tumour growth and survival, mice were equally divided into four groups of (a) normal saline s.c. at 50 μl and 0.5% methylcellulose at 100 μl by gavage (control); (b) morphine sulphate s.c. at 0.714 mg per kg per mouse per day for 7 days and 1 mg per kg per mouse per day thereafter (equivalent to 50 and 75 mg of morphine per day for a 70 kg human, respectively) and 0.5% methylcellulose at 100 μl by gavage to ensure similar administration of vehicle in each animal; (c) celecoxib at 30 mg per kg per mouse by gavage in 0.5% methylcellulose and (d) morphine sulphate plus celecoxib as described in b and c.
To determine the effect of high-dose celecoxib, mice were treated with (a) normal saline s.c. at 50 μl and 0.5% methylcellulose at 100 μl by gavage; (b) morphine sulphate s.c. at 0.714 mg per kg per mouse per day for 7 days and 1 mg per kg per mouse per day thereafter; (c) celecoxib at 100 mg per kg per mouse by gavage in 0.5% methylcellulose and (d) morphine sulphate plus celecoxib as described in IIb and IIc.
Tumour cryosections were immunostained with the endothelial cell marker anti-CD31-PE (1 : 50 dilution; Pharmingen, San Diego, CA, USA) and the nuclear stain DAPI (Molecular Probes, Eugene, OR, USA), as described previously (Poonawala et al, 2005). Three different sections of tumour were selected for image analysis, and digital images of at least three different areas of each section were binarised and linearized to quantify total PE-positive pixels in each image as a measure of such angiogenic parameters as blood vessel length, ends and nodes, using the Image Processing Tool kit, Plug-in Functions for Adobe PhotoShop (Reindeer Games, Asheville, NC, USA), as described before (Gupta et al, 2002).
Thermal escape latency
Tumour-bearing mice were acclimated on the glass surface of the chamber of the Paw Thermal Stimulator System (UCSD, San Diego, CA, USA) for 60 min before the test (Hargreaves et al, 1988). A radiant heat stimulus was applied to the plantar surface of a single hind paw from underneath the glass floor with a projector lamp bulb (CXL/CXR, 8 V, 50 W). Paw withdrawal latency (PWL) to the nearest 0.1 s was automatically recorded when a mouse withdrew its paw from the stimulus. Stimulus intensity was adjusted to derive an average baseline PWL of approximately 10.0 s in PBS-treated control mice. A 20-s stimulus cutoff was used to prevent tissue damage.
Western blot analysis
Whole tumour lysates containing 50 μg of protein were resolved on 3–15% gradient SDS-PAGE gels, followed by transfer to the PVDF membrane (Immobilion, Millipore, Bedford, MA, USA). Membranes were blocked for 1 h at room temperature in 5% non-fat dry milk in Tris-buffered saline with 0.1% Tween 20, and incubated with 1 : 500 dilution of anti-COX-2 antibody (Cayman Chemical, Ann Arbor, MI, USA) or anti-β-actin (Santa Cruz Biotechnology Inc., Santa Cruz, CA) at 4°C overnight. After washing, membranes were incubated with 2° species-specific antibodies conjugated to alkaline phosphatase for 45 min at room temperature. Proteins were visualised with an ECF Western blotting system (Amersham Life Sciences, Buckinghamshire, UK). Chemiluminescent signals were acquired using a Storm 860 PhosphorImager (Molecular Dynamics, Sunnyvale, CA, USA).
Total RNA was isolated from tumours using Trizol reagent (Farooqui et al, 2006). Five micrograms of total RNA was reverse transcribed using a First-Strand Synthesis System (Invitrogen, Carlsbad, CA, USA). PCRs were performed by using Taq DNA polymerase (Continental Lab Products, San Diego, CA, USA). Sequences of primers homologous to the coding region of mouse COX-2 (GenBank acc. no. NM_011198) were 5′-ACTCACTCAGTTTGTTGAGTCATT-3′ (sense nucleotides 1312–1335) and 5′-TTTGATTAGTACTGTAGGGTTAAT-3′ (antisense nucleotides 1871–1894); sequences for mouse GAPDH (GenBank acc. no. NM_001001303) were: 5′-CGTCTTCACCACCATGGAGA-3′ (sense nucleotides 353–371) and 5′-CGGCCATCACGCCACAGTTT-3′ (antisense nucleotides 635–651) (Lim et al, 2006). Amplification was performed for 30 cycles at 94°C for 50 s, 56°C for 50 s and 72°C for 50 s with a final extension cycle for 10 min at 72°C in a PTC-100 thermocycler (MJ Research, Waltham, MA, USA). DNA samples were separated by electrophoresis in 2% agarose gel. The purified PCR products obtained were sequenced (Microchemical Facility, University of Minnesota) to verify that they matched the expected DNA sequences.
Prostaglandin E2 assay
Prostaglandin E2 levels were estimated in whole tumour lysates using an EIA Kit (Cayman Chemical Company, Ann Arbor, MI, USA) as per the manufacturer's instructions. Lysates were purified using acetone precipitation of proteins, and supernatants from each sample were quantified using two different dilutions of each. In this method, the intensity of the colour produced is measured at 412 nm and is inversely proportional to the amount of PGE2 present in the well. Recovery of PGE2 was calculated based on the recovery of a cold spike of PGE2 added to the lysis buffer and processed in parallel with tumour lysates. Prostaglandin E2 content was calculated using a calibration curve prepared with 10–1000 pg per ml of PGE2.
Lungs were fixed in 10% formalin overnight, and paraffin-embedded lung sections were stained with haematoxylin and eosin. Metastatic foci were visualised in a double-blinded manner using a light microscope.
Duration of survival in the four treatment groups was analysed with Kaplan–Meier curves. Equality of survival curves was assessed using the Wilcoxon's log-rank test, which weights differences in the curves by the number of surviving mice (Klein and Moeschberger, 1997).
Tumour weight was analysed with an ANOVA model, and pairwise comparisons among the four treatment groups were made using Tukey's studentised range test (HSD) (Christensen, 2002).
Four measures of vascular complexity (pixels, nodes, ends and line length) were performed at three sites in three mice from each treatment group. Mice were randomly selected from among those living at the end of 14 days. Differences between treatment groups of these measurements were assessed using an ANOVA model with a random intercept for each mouse to account for within-mouse correlation of measurements. Tests for treatment group effects used the between-within method for determining the degrees of freedom and maximum likelihood to fit the model (Fitzmaurice et al, 2004).
We assessed longitudinal changes in pain using a repeated-measures ANOVA model. The model includes three fixed-effect terms of treatment group, day, and treatment group by day interaction. It also includes a random-effect term for the mouse-specific baseline. Tests for treatment group effects at each day used the between-within method for determining the degrees of freedom and maximum likelihood to fit the model.
Morphine upregulates COX-2 expression and PGE2 in tumours
To determine whether morphine treatment exaggerates the expression of COX-2, we evaluated COX-2 protein and mRNA expression in SCK mammary tumours in A/J mice after 13 days of treatment using Protocol I. Cyclooxygenase-2 protein and mRNA levels were increased by five- to seven-fold after morphine treatment as compared to PBS-treated mice (Figure 1A and B). Upregulation of COX-2 expression was higher in endothelial and tumour cells in morphine-treated mice compared to PBS-treated controls, based on immunofluorescent staining of tumour cryosections (Figure 1C). Notably, morphine induced a much stronger expression of COX-2 in tumour cells than in endothelial cells. The upregulation of COX-2 protein and gene expression was accompanied by an ∼5-fold increase in PGE2 (P<0.05 vs control) in tumours of mice treated with morphine (Figure 1D). Co-administration of celecoxib blocked this morphine-induced increase in COX-2 expression and PGE2.
Celecoxib inhibits morphine-induced tumour angiogenesis
We next examined the effect of combined treatment with morphine and celecoxib on tumour angiogenesis in tumour-bearing mice. Immunostained sections of SCK tumours from mice treated with morphine for 2 weeks showed an increase in angiogenesis (Figure 2A). Tumours in morphine-treated mice appeared to have a higher density of dilated and branching vessels as compared to PBS-treated controls. In contrast, tumours from mice treated with both celecoxib and morphine appeared to have a lower vessel density with unevenly distributed and thinner vessels, among which were interspersed hollow spaces suggestive of the absence of endothelial and tumour cells in these areas. Similarly, tumours in mice treated with celecoxib alone had small, stringy vessels with large unstained areas suggestive of the absence of endothelial and tumour cells. These qualitative observations of immunostained tumour sections were corroborated by quantitative morphometric analysis (Figure 2B). We observed increased microvessel density (CD31-PE-positive pixels, 1.4-fold), vessel number (ends, 1.9-fold), total length (1.4-fold) and branching (nodes, 2.2-fold) in morphine-treated mice as compared to PBS-treated controls (P=0.001 vs control for all values). Combined treatment of celecoxib with morphine significantly reduced all angiogenic parameters as compared to morphine by itself. Tumours in the celecoxib-treated group had lowered vessel density, number, length and branching as compared to controls, but there was no statistically significant difference. Taken together, these data suggest that morphine stimulates tumour angiogenesis in SCK tumours similar to that shown for MCF7 human tumours in nude mice (Gupta et al, 2002), whereas celecoxib inhibits morphine-induced angiogenesis. Importantly, celecoxib blocked morphine-induced neovascularisation in these tumours, suggesting that COX-2 plays a critical role in the tumour-promoting activity of morphine.
Celecoxib inhibits morphine-induced tumour growth and metastasis
We next evaluated the effect of morphine and celecoxib on tumour growth in tumour-bearing mice. After 13 days of treatment with morphine, mean tumour weight in morphine-treated mice was 1.35 g compared to 0.95 g in PBS-treated controls (P<0.001; Figure 2C). Celecoxib significantly reduced morphine-induced tumour growth resulting in a mean tumour weight of 0.69 g compared to PBS-treated controls (P<0.001). Furthermore, celecoxib-treated mice showed no significant difference in their tumour weights as compared to PBS-treated controls after 13 days of treatment with celecoxib. Interestingly, tumours in morphine and celecoxib co-administered mice were significantly smaller than PBS-treated mice (P<0.001).
Increased angiogenesis and tumour growth after 13 days of morphine treatment was accompanied by increased metastasis (Table 1). A significant number of mice died on day 14 due to the invasiveness of this tumour model (discussed below). Therefore, we examined metastases in the lungs of mice that died on day 14 and the remaining mice that were alive after euthanasia. One hundred percent of morphine-treated mice showed metastatic foci in the lungs as compared to only 70% of control mice with lung metastases (Table 1). Lungs recovered from dead mice showed metastases in all groups except celecoxib. Of the surviving mice, 100% of morphine-treated mice showed lung metastases as compared to only 57% of PBS-treated mice. In contrast, celecoxib alone or co-administered with morphine showed metastasis in only 30 or 25% surviving mice, respectively. From these data, it appears that celecoxib prevents metastases in this model and also that induced by morphine.
Celecoxib plus morphine increases survival
We examined the effect of morphine and celecoxib on survival in tumour-bearing mice. Treatments were started 1 day after tumour cell injection and continued for 14 days. The day of death represents the number of days after the initiation of treatments. Morphine-treated mice had the worse survival rate with 70% dying by day 14 as compared to 30% of PBS-treated controls (Figure 3A). Only 20% of mice treated with morphine plus celecoxib had died by day 14, with only one death occurring by day 13. Surprisingly, 20% of mice treated with celecoxib died as early as day 4 and 40% had died by day 9, while the remainder of the mice survived until day 13 and 50% had died by day 14.
In the SCK tumour model, tumours at day 4 are barely palpable and non-metastatic. Therefore, the early death of mice after celecoxib treatment is not due to cancer progression. Several studies on rodents have used celecoxib in the range 30–250 mg per kg per mouse for several days to weeks (Koki and Masferrer, 2002; Leahy et al, 2002). None of these studies reported toxicity or adverse effects of these COX-2 inhibitors on the survival of mice. The only difference between our study and others is that our study used A/J mice and a SCK tumour model. It is possible that A/J mice are more sensitive to COX-2 toxicity.
To address the toxicity induced by celecoxib, we performed the same survival experiments in tumour-bearing mice with an increased initial dose of celecoxib 100 mg per kg per day on day 1 per protocol II. Surprisingly, 5 out of 10 mice receiving 100 mg per kg of celecoxib died within 24 h, and the other 5 appeared moribund (Table 2). After unsuccessful attempts to revive them, the remaining mice were euthanised after 48 h. However, mice that were co-administered morphine with 100 mg per kg of celecoxib did not show any signs of toxicity and appeared to have the same levels of activity as morphine-treated mice after 24 h. After the first day, the dose of celecoxib was reduced to 30 mg per kg per day with continued administration of 1 mg per kg per day morphine, and by day 10 of treatment (11 days after tumour cell injection), 100% of the mice were still alive. Thus, celecoxib alone impaired survival via non-cancer-related mechanisms. In contrast, morphine at higher doses was not toxic but still impaired survival at later stages by exaggerating tumour growth and metastasis. When administered together even at higher doses, morphine and celecoxib blocked the adverse effect of the other, resulting in increased survival. Taken together, these data show that a combination of morphine and celecoxib increases the survival rate of tumour-bearing mice as compared to treatment with either morphine or celecoxib alone, which is consistent with our findings that combination treatment also decreases angiogenesis and tumour burden.
Effect of celecoxib on morphine-induced analgesia
Treatment effects on pain were tested using the PWL of the tumour-bearing leg, using a modified Hargreaves device. Baseline PWL was determined on day 0 before starting the treatments and 24 h after injecting the tumour cells. This time was chosen to avoid recording effects of pain due to the injection procedure and inflammation that could occur immediately after injection. Mice had no signs of swelling, inflammation or impaired mobility 24 h after tumour-cell injections. Baseline PWLs were the same for all treatment groups. Paw withdrawal latencies were significantly decreased in mice after 5 days of treatments (or 6 days after tumour-cell injection) with PBS and celecoxib (P<0.0001 vs baseline at day 0); both groups had palpable and measurable tumours on day 5. Morphine alone had an anti-nociceptive effect after 5 days of treatment (P<0.003 vs PBS-treated control), but had no effect after 10 and 14 days of treatment as compared to controls. In contrast to the effect of celecoxib or morphine alone, the co-administration of both resulted into a continuous analgesic effect for the entire 14 days of treatment. Paw withdrawal latencies in this group were no different than the baseline throughout the 14-day period. The duration of heat tolerance was significantly higher when both drugs were co-administered as compared to the effect of all other treatment groups (P<0.0001 on days 10 and 14 for morphine+celecoxib vs morphine or celecoxib or control). Although celecoxib treatment resulted in decreased latency vs control on day 10 (P<0.003), there was no statistically significant difference between morphine and control on day 10. On day 14, PWLs decreased further after morphine or control treatment as compared to day 10, and were no different than celecoxib treatment on day 14. Injections during the course of treatment for all groups did not result in any noticeable abnormal behaviour.
Our study provides a proof of principle that the chronic use of opioids stimulates COX-2, leading to increased prostaglandin(s), impaired analgesia and increased tumour angiogenesis, growth, metastasis and mortality. We show that inhibition of COX-2 by co-administration of celecoxib prevents morphine-induced tumour growth and metastases and increases survival. Furthermore, co-administration of celecoxib with morphine provides analgesia even after 14 days of treatment, suggesting that celecoxib may have a salutary effect on opioid analgesia. Cyclooxygenase-2 inhibition may at least partly explain the clinical improvements in postoperative pain control that have been seen with a combination of opioids and COX-2 inhibitors and/or NSAIDs (Kroin et al, 2002; Malan et al, 2003).
Chronic administration of opioids has been linked to angiogenesis, progression of tumour growth and hyperalgesia. Tolerance to the analgesic effect of opioids is believed to be mediated at least partly by prostaglandin-mediated mechanisms (Powell et al, 1999). Intrathecal administration of COX-2 inhibitors attenuates the development of tolerance to morphine (Wong et al, 2000). Further, intrathecally administered COX-2 inhibitors have significant analgesic activity in hyperalgesic states not associated with inflammation, possibly by modulation of COX-2 that is constitutively expressed in the spinal cord (Svensson and Yaksh, 2002). Increased PGE2 leads to thermal hyperalgesia in the spinal cord following intrathecal delivery of protease-activated receptor (PAR)-derived peptide (Koetzner et al, 2004). Although the mechanisms of morphine and PAR-induced activation of PGE2 may be different, it appears that PGE2 mediates thermal hyperalgesia, irrespective of the central or peripheral site of action. Furthermore, inhibition of morphine-induced PGE2 by celecoxib in tumours is accompanied by prevention of thermal nociception in morphine-treated mice. It has been suggested before that nerve injury and TNF-induced pain-related behaviour is partly dependent upon peripheral prostaglandins (Schafers et al, 2004). Morphine stimulates TNFα secretion both centrally and peripherally (Gupta and Stephenson, 2007). It is therefore possible that morphine-induced upregulation of COX-2 and PGE2 may be due to the elevation of TNFα caused by morphine and/or due to some other mechanism.
The promotion of tumour growth by morphine appears to be dependent on PGE2-mediated stimulation of angiogenesis. Morphine-induced upregulation of COX-2 is critical in the progression of tumour angiogenesis, because tumour cell-derived COX-2 profoundly influences angiogenesis (Chang et al, 2004). Although endothelial COX-2 does not appear to be extremely upregulated, increased COX-2 in non-endothelial cells appears to be responsible for increased PGE2 in tumours of morphine-treated mice. Koki and co-workers have proposed that COX-2 is upregulated in tumour epithelial cells and stroma in response to growth factors leading to increased production of PGE2 that acts on endothelium and promotes angiogenesis (Koki and Masferrer, 2002). Similarly, morphine treatment stimulates COX-2 in non-endothelial cells in the tumour, which leads to increased PGE2 that then acts on endothelium to stimulate tumour angiogenesis. It is likely that celecoxib inhibits COX-2 in tumour cells, resulting in the inhibition of tumour growth and decreased angiogenesis. Therefore, celecoxib may inhibit angiogenesis indirectly via inhibition of tumour growth, rather than directly via anti-angiogenic mechanisms. This is further supported by observations of Koki and Masferrer (2002) that nude mice showed a decline in mammary tumour growth continuously throughout a 3-week treatment with celecoxib.
These anti-angiogenic and anti-tumour effect in rodents have been observed using extremely high doses of celecoxib. However, survival of mice was not impaired, even though animals were fed chow supplemented with extremely high amounts of celecoxib (1600 p.p.m.; ∼250 mg per kg per day) (Koki and Masferrer, 2002). Similarly, Leahy et al (2002) reported that nude mice treated for 40 days with celecoxib (25 mg per kg per day) had a significant reduction in tumour growth of HT-29 and HCT-116 human colon carcinoma xenografts and a reduction in the proliferation of microvascular endothelial cells). In the same study, rats implanted with pellets containing FGF2 in an intrasomal pocket in the cornea and treated with celecoxib 30 mg per kg per day for 4 or 6 days by gavage showed a significant reduction in corneal neovascularisation. In contrast to these observations, we found that mice treated with celecoxib at 30 mg per kg per day started dying 4 days after treatment. After 7 days of treatment, only 60% of the celecoxib-treated mice survived, whereas 100% of mice were still surviving in the group treated with celecoxib plus morphine. This early mortality in the celecoxib-treated group was not due to tumour growth or metastases. Administration of a higher dose of celecoxib (100 mg per kg per day) resulted in an even higher early mortality rate (50% within the first 24 h). Impaired survival in celecoxib-treated mice was therefore likely due to drug toxicity. Importantly, at both the low and high doses, celecoxib co-administered with morphine did not impair mouse survival. The survival rate in mice treated with celecoxib plus morphine was comparable to PBS-treated mice, and this rate was significantly better compared to morphine-only treatment. We believe that high doses of celecoxib used by us and others may have nonspecific activity beyond selectively inhibiting COX-2 activity. We used high doses of celecoxib, because (a) the studies described above show an inhibition of angiogenesis with the doses we used and (b) to examine if morphine could prevent high-dose celecoxib-induced toxicity. In our study, mice treated with high doses (100 mg per kg) of celecoxib and morphine survived, whereas those treated with a high dose of celecoxib alone died within 24 h of treatment, suggesting that morphine prevents high-dose celecoxib toxicity.
The loss of mice 13 days after morphine treatment appears to be related to increased tumour growth and metastasis as compared to PBS-treated mice. In contrast, mice treated with celecoxib plus morphine showed significantly reduced tumour growth as compared to PBS or celecoxib-treated mice. Immunofluorescent images of tumour sections show appreciably wider vessels in both groups of mice treated with morphine or morphine plus celecoxib. As morphine appears to induce tumour vessel dilation, we speculate that it may facilitate the availability of celecoxib in the ischaemic tumour. Earlier observations have shown that morphine induces vasodilatation by stimulating NO (Stefano et al, 1995). In addition, recent observations from our laboratory show that 3 weeks of morphine treatment stimulates the expression of the vasodilatory enzymes iNOS, eNOS, COX-2 and HO1 in the kidneys of C3H mice bearing NCTC 2472 tumours (Arerangaiah et al, 2007). Upregulation of NO by morphine may also compensate for the vasoconstrictive and hypertensive effect of celecoxib on cardiovascular haemodynamics. As COX-2 activity and NOS activity co-stimulate each other, it is likely that inhibition of COX-2 reduces NO production, resulting in haemodynamic insult. Addition of morphine may therefore prevent celecoxib toxicity by activating alternative pathways, including NOS and HO1. Irrespective of the mechanism, our data show that co-administration of celecoxib with morphine reduces tumour metastasis induced by morphine. Based on these data and supportive observations, we conclude that celecoxib and morphine may directly and/or indirectly block each other's ‘ill effects’. Therefore, although celecoxib reduces morphine-induced angiogenesis, tumour growth and metastasis, morphine prevents celecoxib-induced mortality and both together provide a better analgesic response and improved survival. It is important to note that co-administration of celecoxib with morphine resulted in significantly reduced tumour growth as well as pain relief as compared to control or celecoxib or morphine alone. Our observations are supported by another study showing that stress and morphine together reduced survival of rats bearing MAT13762B mammary ascites tumour, which was reversed by naltrexone, suggesting an opiate receptor-mediated mechanism and reduced NK cell activity (Lewis et al, 1983, 1984). Thus, improved survival in mice treated with a combination of morphine and celecoxib may be due to better analgesia in addition to the inhibition of angiogenesis and tumour growth.
Although our study shows that clinically relevant doses of morphine stimulate tumour angiogenesis and tumour growth and lead to reduced survival, some past studies have observed conflicting effects of morphine on angiogenesis and tumour growth. We believe that these differences are primarily due to the extremely high cytotoxic doses of morphine given in these studies as compared to clinically and physiologically relevant and opioid receptor-specific doses that ought to be administered. Ishikawa et al (1993) showed that 10 mg per kg morphine increases EL-4 leukaemia in C57/BL6 mice and Sarcoma 180 carcinoma in ddY mice in tumour growth in mice after 10 days of treatment. In this systematically performed investigation, morphine did not stimulate the growth of EL-4, P388, MM-46 or Meth-A cells in vitro. These data are in agreement with our earlier studies showing that morphine does not have any effect on MCF-7 human breast cancer cell proliferation (Farooqui et al, 2006) and that morphine stimulates MCF-7 breast tumour growth by promoting angiogenesis (Gupta et al, 2002). Morphine can affect tumour growth by (a) directly acting on the tumour cells, (b) by directly acting on the endothelial cells and (c) by acting on the central nervous system-mediated secretion of cytokines and growth factors that may alter the tumour microenvironment (Gupta and Stephenson, 2007). Morphine acts on the endothelium in a paracrine manner by increasing COX-2 activity in the tumour cells that in turn promotes tumour angiogenesis by increased PGE2. Therefore, morphine by itself does not modulate tumour cell growth in culture but, due to upregulation of COX-2 in tumour cells, stimulates angiogenesis in vivo, resulting in increased tumour growth and metastasis. These observations suggest that in vivo examination may be more relevant than studying the effect of morphine on isolated cells.
Conversely, some studies suggest an inhibitory effect of morphine on tumour growth. A high dose of morphine was shown to inhibit tumour growth and lung metastasis in a model of cancer pain in B16-BL6 melanoma cell growth in C57/BL6 mice, and to inhibit the adhesion, invasion and metastasis of colon cancer cells in vitro (Harimaya et al, 2002; Sasamura et al, 2002). Similarly, another study showed that high doses of morphine (up to 30 mg per kg) as well as naloxone inhibit MCF-7 breast cancer after d22 and MDA MB231 breast cancer progression after d14 in nude mice (last periods of observation) (Tegeder et al, 2003). However, the doses of morphine used in this study are not clinically relevant because humans suffering from cancer are administered far lower doses of morphine than those required for analgesia in rodents. Moreover, in MCF7 tumour model, measurable tumour growth is observed only after 2 weeks of tumour cell xenografts and differences between treatments are observed after 3 weeks of treatment, but the study was concluded on day 22. Thus, the use of clinically toxic doses of morphine in this study, as well as its early conclusion, may account for the observed inhibition of tumour growth.
Our observation of morphine-induced tumour growth is supported by earlier studies showing an inhibition of tumour growth in rodents with low dose of the OR antagonists naloxone and naltrexone, but not with higher dose (Zagon and McLaughlin, 1983a, 1983b). At low doses, naloxone antagonizes OR activity, but at higher doses, it may have nonspecific effects. More recently, the OR antagonists naloxone and naltrexone were shown to inhibit breast cancer growth in mice and rats, respectively (Koo et al, 1996; Tegeder et al, 2003). These data suggest that tumour progression may be mediated by endogenous opioids and opioids receptors acting directly on tumour cells or endothelial cells, or by other paracrine and centrally mediated mechanisms.
Some studies have demonstrated inhibition of angiogenesis by high doses of morphine and β-endorphin in a CAM assay (Pasi et al, 1991). However, these results may be due to the nonspecific activity of excessive morphine dosage, and the difference in OR presentation in the embryonic vs tumour endothelium. OR activity is believed to be upregulated in the tumour endothelium because of the high concentrations of pro-inflammatory cytokines and growth factors in the tumour microenvironment. Indeed, these cytokines and growth factors are known to upregulate the expression of ORs in several cell types (Gupta chapter endothelium, 2007), and growth factors such as VEGF have been shown to upregulate MOR expression in endothelial cells (Chen et al, 2006). As such, we believe that these findings do not contradict our hypothesis or results, but rather point to the complications of using morphine at excessive doses and in environments dissimilar to the tumour endothelium.
Our observations are in agreement with a study by Singleton et al (2006) that observed morphine-induced angiogenesis, using physiologically relevant doses of morphine. The authors showed that clinically relevant doses of morphine and MOR-specific agonist DAMGO stimulate angiogenesis by transactivating VEGFR2. The MOR-specific antagonist MNTX antagonized morphine-induced angiogenesis. Taken together, these observations suggest that excessive doses of morphine may inhibit angiogenesis because of their nonspecific cytotoxicity, whereas physiologically relevant doses stimulate angiogenesis via OR-mediated activity.
In conclusion, pain is closely linked to disease progression and metastasis in cancer. Therefore, finding suitable strategies to combat both without having a deleterious effect on one or the other is necessary in order to provide maximum benefit to the patient, resulting in better quality of life and improved survival. Considering that opioids are the only choice of treatment for severe pain, the addition of celecoxib to cancer-related pain management regimens may improve the analgesic response to morphine and also prevent morphine-induced angiogenesis, tumour growth and metastasis. Therefore, the hypothesis that has been proved in mice in this study warrants human trial to examine its application in the management of pain in cancer patients.
Arerangaiah R, Chalasani N, Udager AM, Weber ML, Manivel JC, Griffin RJ, Song CW, Gupta K (2007) Opioids induce renal abnormalities in tumor-bearing mice. Nephron Exp Nephrol 105: e80–e89
Birnbaum Y, Ye Y, Rosanio S, Tavackoli S, Hu ZY, Schwarz ER, Uretsky BF (2005) Prostaglandins mediate the cardioprotective effects of atorvastatin against ischemia–reperfusion injury. Cardiovasc Res 65: 345–355
Chang SH, Liu CH, Conway R, Han DK, Nithipatikom K, Trifan OC, Lane TF, Hla T (2004) Role of prostaglandin e2-dependent angiogenic switch in cyclooxygenase 2-induced breast cancer progression. Proc Natl Acad Sci USA 101: 591–596
Chen C, Farooqui M, Gupta K (2006) Morphine stimulates vascular endothelial growth factor-like signaling in mouse retinal endothelial cells. Curr Neurovasc Res 3: 171–180
Christensen R (2002) Plane Answers to Complex Questions. The Theory of Linear Models. 3rd edn, New York: Springer
Farooqui M, Geng ZH, Stephenson EJ, Yee D, Gupta K (2006) Naloxone acts as an antagonist of estrogen receptor in MCF7 cancer cells. Mol Cancer Ther 5: 611–620
Fitzmaurice G, Laird N, Ware J (2004) Applied Longitudinal Analysis. Hoboken, New Jersey: John Wiley & Sons, Inc
Griffin RJ, Williams BW, Wild R, Cherrington JM, Park H, Song CW (2002) Simultaneous inhibition of the receptor kinase activity of vascular endothelial, fibroblast, and platelet-derived growth factors suppresses tumor growth and enhances tumor radiation response. Cancer Res 62: 1702–1706
Gupta K, Kshirsagar S, Chang L, Schwartz R, Law PY, Yee D, Hebbel RP (2002) Morphine stimulates angiogenesis by activating proangiogenic and survival-promoting signaling and promotes breast tumor growth. Cancer Res 62: 4491–4498
Gupta K, Stephenson EJ (2007) Opioid receptors in endothelium. In Endothelial Biomedicine, William C Aird (Ed), pp 451–460. New York, NY: Cambridge University Press
Hargreaves K, Dubner R, Brown F, Flores C, Joris J (1988) A new and sensitive method for measuring thermal nociception in cutaneous hyperalgesia. Pain 32: 77–88
Harimaya Y, Koizumi K, Andoh T, Nojima H, Kuraishi Y, Saiki I (2002) Potential ability of morphine to inhibit the adhesion, invasion and metastasis of metastatic colon 26-L5 carcinoma cells. Cancer Lett 187: 121–127
Ishikawa M, Tanno K, Kano A, Takayanagi Y, Sasaki K (1993) Enhancement of tumor growth by morphine and its possible mechanism in mice. Biol Pharm Bull 16: 762–766
Julius D, Basbaum AI (2001) Molecular mechanisms of nociception. Nature 413: 203–210
Klein JP, Moeschberger ML (1997) Survival Analysis Techniques for Censored and Truncated Data. New York: Springer
Koetzner L, Gregory JA, Yaksh TL (2004) Intrathecal protease-activated receptor stimulation produces thermal hyperalgesia through spinal cyclooxygenase activity. J Pharmacol Exp Ther 311: 356–363
Koki AT, Masferrer JL (2002) Celecoxib: a specific cox-2 inhibitor with anticancer properties. Cancer Control 9: 28–35
Koo KL, Tejwani GA, Abou-Issa H (1996) Relative efficacy of the opioid antagonist, naltrexone, on the initiation and promotion phases of rat mammary carcinogenesis. Anticancer Res 16: 1893–1898
Kroin JS, Buvanendran A, McCarthy RJ, Hemmati H, Tuman KJ (2002) Cyclooxygenase-2 inhibition potentiates morphine antinociception at the spinal level in a postoperative pain model. Reg Anesth Pain Med 27: 451–455
Leahy KM, Ornberg RL, Wang Y, Zweifel BS, Koki AT, Masferrer JL (2002) Cyclooxygenase-2 inhibition by celecoxib reduces proliferation and induces apoptosis in angiogenic endothelial cells in vivo. Cancer Res 62: 625–631
Levy MH (1996) Pharmacologic treatment of cancer pain. N Engl J Med 335: 1124–1132
Lewis JB, Shavit Y, Tennan GW, Gale RP, Liebskind JC (1983–1984) Stress and morphine affect survival of rats challenged with a mammary ascites tumor (MAT13762B). Nat Immun Cell Growth Regul 3: 43–50
Lim H, Son KH, Chang HW, Kang SS, Kim HP (2006) Effects of anti-inflammatory biflavonoid, ginkgetin, on chronic skin inflammation. Biol Pharm Bull 29: 1046–1049
Malan Jr TP, Marsh G, Hakki SI, Grossman E, Traylor L, Hubbard RC (2003) Parecoxib sodium, a parenteral cyclooxygenase 2 selective inhibitor, improves morphine analgesia and is opioid-sparing following total hip arthroplasty. Anesthesiology 98: 950–956
Malmberg AB, Yaksh TL (1992) Hyperalgesia mediated by spinal glutamate or substance p receptor blocked by spinal cyclooxygenase inhibition. Science 257: 1276–1279
Mantyh PW (2006) Cancer pain and its impact on diagnosis, survival and quality of life. Nat Rev Neurosci 7: 797–809
Nedelec E, Abid A, Cipolletta C, Presle N, Terlain B, Netter P, Jouzeau J (2001) Stimulation of cyclooxygenase-2-activity by nitric oxide-derived species in rat chondrocyte: Lack of contribution to loss of cartilage anabolism. Biochem Pharmacol 61: 965–978
Pasi A, Qu BX, Steiner R, Senn HJ, Bar W, Messiha FS (1991) Angiogenesis: modulation with opioids. Gen Pharmacol 22: 1077–1079
Poonawala T, Levay-Young BK, Hebbel RP, Gupta K (2005) Opioids heal ischemic wounds in the rat. Wound Repair Regen 13: 165–174
Powell KJ, Hosokawa A, Bell A, Sutak M, Milne B, Quirion R, Jhamandas K (1999) Comparative effects of cyclo-oxygenase and nitric oxide synthase inhibition on the development and reversal of spinal opioid tolerance. Br J Pharmacol 127: 631–644
Salvemini D, Misko TP, Masferrer JL, Seibert K, Currie MG, Needleman P (1993) Nitric oxide activates cyclooxygenase enzymes. Proc Natl Acad Sci USA 90: 7240–7244
Salvemini D, Seibert K, Masferrer JL, Misko TP, Currie MG, Needleman P (1994) Endogenous nitric oxide enhances prostaglandin production in a model of renal inflammation. J Clin Invest 93: 1940–1947
Samad TA, Moore KA, Sapirstein A, Billet S, Allchorne A, Poole S, Bonventre JV, Woolf CJ (2001) Interleukin-1beta-mediated induction of cox-2 in the CNS contributes to inflammatory pain hypersensitivity. Nature 410: 471–475
Sasamura T, Nakamura S, Iida Y, Fujii H, Murata J, Saiki I, Nojima H, Kuraishi Y (2002) Morphine analgesia suppresses tumor growth and metastasis in a mouse model of cancer pain produced by orthotopic tumor inoculation. Eur J Pharmacol 441: 185–191
Schafers M, Marziniak M, Sorkin LS, Yaksh TL, Sommer C (2004) Cyclooxygenase inhibition in nerve-injury- and TNF-induced hyperalgesia in the rat. Exp Neurol 185: 160–168
Singleton PA, Lingen MW, Fekete MJ, Garcia JG, Moss J (2006) Methylnaltrexone inhibits opiate and VEGF-induced angiogenesis: role of receptor transactivation. Microvasc Res 72: 3–11
Stefano GB, Hartman A, Bilfinger TV, Magazine HI, Liu Y, Casares F, Goligorsky MS (1995) Presence of the mu3 opiate receptor in endothelial cells. Coupling to nitric oxide production and vasodilation. J Biol Chem 270: 30290–30293
Svensson CI, Yaksh TL (2002) The spinal phospholipase-cyclooxygenase-prostanoid cascade in nociceptive processing. Annu Rev Pharmacol Toxicol 42: 553–583
Tegeder I, Grosch S, Schmidtko A, Haussler A, Schmidt H, Niederberger E, Scholich K, Geisslinger G (2003) G protein-independent G1 cell cycle block and apoptosis with morphine in adenocarcinoma cells: involvement of p53 phosphorylation. Cancer Res 63: 1846–1852
Wong C-S, Hsu M-M, Chou R, Chou YY, Tung CS (2000) Intrathecal cyclooxygenase inhibitor administration attenuates morphine antinociceptive tolerance in rats. Brit J Anaesth 85: 747–751
Zagon IS, McLaughlin PJ (1983a) Naltrexone modulates tumor response in mice with neuroblastoma. Science 221: 671–673
Zagon IS, McLaughlin PJ (1983b) Opioid antagonists inhibit the growth of metastatic murine neuroblastoma. Cancer Lett 21: 89–94
We are thankful to Mihir Gupta, Pankaj Gupta, MD and Michael J Franklin for critical review of the paper, to Ms Carol Taubert for preparation of the document, and to Brent W Williams for technical assistance. This work is supported by NIH Grants HL68802, CA109582 and the Susan G Komen Breast Cancer Foundation (to KG) and CA109582 (to RJG).
Rights and permissions
From twelve months after its original publication, this work is licensed under the Creative Commons Attribution-NonCommercial-Share Alike 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/
About this article
Cite this article
Farooqui, M., Li, Y., Rogers, T. et al. COX-2 inhibitor celecoxib prevents chronic morphine-induced promotion of angiogenesis, tumour growth, metastasis and mortality, without compromising analgesia. Br J Cancer 97, 1523–1531 (2007). https://doi.org/10.1038/sj.bjc.6604057
- breast cancer
This article is cited by
Kinin B1 and B2 receptors mediate cancer pain associated with both the tumor and oncology therapy using aromatase inhibitors
Scientific Reports (2023)
Effect of single-shot intrathecal morphine versus continuous epidural analgesia on length of stay after gastrectomy for cancer: a retrospective cohort study
Gastric Cancer (2023)
Effect of aspirin on PET parameters in primary non-small cell lung cancer and its relationship with prognosis
BMC Cancer (2020)
Opioids and Cancer Mortality
Current Treatment Options in Oncology (2020)
The dual effect of morphine on tumor development
Clinical and Translational Oncology (2019)