Electroacupuncture for Control of Myeloablative Chemotherapy–Induced Emesis a Randomized Controlled Trial

CLINICAL INVESTIGATION

Electroacupuncture for Control of Myeloablative Chemotherapy–Induced Emesis A Randomized Controlled Trial

Joannie Shen, MD, MPH Context High-dose chemotherapy poses considerable challenges to emesis man- Neil Wenger, MD, MPH agement. Although prior studies suggest that acupuncture may reduce nausea and emesis, it is unclear whether such benefit comes from the nonspecific effects of at- John Glaspy, MD, MPH tention and clinician-patient interaction. Ron D. Hays, PhD Objective To compare the effectiveness of electroacupuncture vs minimal needling Paul S. Albert, PhD and mock electrical stimulation or antiemetic medications alone in controlling emesis among patients undergoing a highly emetogenic chemotherapy regimen. Christina Choi, OMD Design Three-arm, parallel-group, randomized controlled trial conducted from March Paul G. Shekelle, MD, PhD 1996 to December 1997, with a 5-day study period and a 9-day follow-up. IGH-DOSE, MULTIPLE-DAY, Setting Oncology center at a university medical center. multiple-drug myeloabla- Patients One hundred four women (mean age, 46 years) with high-risk breast cancer. tive chemotherapy poses sub- stantial challenges to emesis Interventions Patients were randomly assigned to receive low-frequency electroacu- control.H The combination of chemo- puncture at classic antiemetic acupuncture points once daily for 5 days (n=37); minimal needling at control points with mock electrostimulation on the same schedule (n=33); therapy agents is highly emetogenic; or no adjunct needling (n=34). All patients received concurrent triple antiemetic phar- most patients have experienced emesis macotherapy and high-dose chemotherapy (cyclophosphamide, cisplatin, and carmustine). with multiple courses of prior chemo- Main Outcome Measures Total number of emesis episodes occurring during the 5-day therapy; and patients may have re- study period and the proportion of emesis-free days, compared among the 3 groups. ceived other medical care or medications and adjuncts that can contribute Results The number of emesis episodes occurring during the 5 days was lower for pa- to emesis. During the last 2 decades, new tients receiving electroacupuncture compared with those receiving minimal needling or pharmacotherapy alone (median number of episodes, 5, 10, and 15, respectively; PϽ.001). effective antiemetic pharmacological The electroacupuncture group had fewer episodes of emesis than the minimal needling agents have helped to improve control group (PϽ.001), whereas the minimal needling group had fewer episodes of emesis than of chemotherapy-induced emesis. Be- the antiemetic pharmacotherapy alone group (P =.01). The differences among groups cause of concerns about pharmacoki- were not significant during the 9-day follow-up period (P=.18). netic interaction between high-dose Conclusions In this study of patients with breast cancer receiving high-dose chemo- chemotherapy agents and the new an- therapy, adjunct electroacupuncture was more effective in controlling emesis than mini- tiemetic medications, some patients re- mal needling or antiemetic pharmacotherapy alone, although the observed effect had ceiving intense multiple-agent, myeloa- limited duration. blative chemotherapy regimens might JAMA. 2000;284:2755-2761 www.jama.com not be able to use these newer antiemetics concurrently.1-3 This constellation of factors makes the management of eme- Author Affiliations: Laboratory of Clinical Studies, Research Branch, National Cancer Institute, National Brain Electrophysiology and Imaging Section, Na- Institutes of Health, Bethesda, Md (Dr Albert); and De- sis difficult. tional Institute of Alcoholism and Alcohol Abuse, Na- partment of Medicine, West-L.A. VAMC, Veterans Af- A recent National Institutes of Health tional Institutes of Health, Bethesda, Md (Dr Shen); fairs Health Services Research and Development Ser- Department of Medicine, Division of General Inter- vice, Los Angeles, Calif (Dr Shekelle). Dr Choi is in Consensus Development Conference nal Medicine and Health Services Research (Drs private practice in Santa Monica, Calif. report concluded that acupuncture was Wenger and Hays) and Department of Medicine, Di- Corresponding Author and Reprints: Joannie Shen, MD, vision of Hematology and Oncology (Dr Glaspy), UCLA MPH, National Institutes of Health, Laboratory of Clini- efficacious in reducing emesis associ- School of Medicine, Los Angeles, Calif; RAND, Santa cal Studies/NIAAA, Room 6 S-240, Mail Stop 1610, 10 ated with chemotherapy.4 However, it Monica, Calif (Drs Hays and Shekelle); Biometric Center Dr, Bldg 10, Bethesda, MD 20892-1610.

Downloaded from www.jama.com on August 23, 2007 ELECTROACUPUNCTURE FOR CHEMOTHERAPY-INDUCED EMESIS was suspected that such benefit was due patients who had brain metastases; life’s motherapy drugs were as follows: cy- to a placebo effect. The one published threatening concurrent nonmalignant clophosphamide, 1875 mg/m2 body- randomized controlled trial that used conditions; active infection, including an surface area, per day over 60 minutes, a sham acupuncture control included active skin infection over the proposed starting at 9 AM for 3 days; cisplatin, 55 10 patients and was limited by its use treatment area; any condition that com- mg/m2 body-surface area, per day with of a crossover design without a wash- promised their ability to give informed continuous infusion, starting at 9 AM for out period.5 Other data that support the consent; or a cardiac pacemaker. A 3 days; and on day 4, carmustine, 600 use of acupuncture to control emesis research assistant conducted a face-to- mg/m2 body-surface area, over 2 hours, include a systematic review and a re- face interview to obtain the following starting at 9 AM, immediately after the cent meta-analysis in the postopera- information: sociodemographic charac- cisplatin dose was completed. tive setting.6,7 However, physiologic rea- teristics, history of chemotherapy, his- Antiemetic Agents. All patients also sons suggest that successful treatment tory of nausea and emesis related to received the same triple pharmacologi- with acupuncture might be different be- motion sickness, morning sickness, cal agents for emesis management. The tween chemotherapy–induced and patients’ expectations about the adverse regimen included prochlorperazine pre- postoperative emesis.8,9 effects of chemotherapy, and the ben- chemotherapy loading, 10 mg in 100 mL In this study, we assessed a standard- efits of antiemetic treatment. of normal saline, intraveneously, over 10 ized electroacupuncture protocol as minutes, followed by continuous intra- an adjunct to antiemetic pharmaco- Randomization Procedure venous infusion at 1 mg/m2 body- therapy for controlling emesis associ- The patients were randomly assigned surface area per hour; lorazepam, 1 ated with intensive, multiple-drug, com- without stratification to receive 1 of 3 mg/m2 body-surface area, intrave- bination myeloablative chemotherapy treatment options: electroacupunc- nously, every 4 hours; and diphenhy- compared with minimal needling or an- ture, minimal needling, or antiemetic dramine hydrochloride, 25 mg/m2 body- tiemetic pharmacotherapy alone. We hy- drugs alone. The 2 interventions were surface area, intravenously, every 6 pothesized a priori that minimal nee- described to the patient as “classical acu- hours. These medications were started dling treatment might have a greater puncture” or “non-classical acupunc- 1 hour prior to chemotherapy and were effect than pharmacotherapy alone be- ture.” We used the term classical acu- continued until 48 hours after the last cause of the nonspecific effects of needles puncture to describe a protocol for chemotherapy infusion. Rescue medi- and the additional attention and care the electroacupuncture at sites that are in- cations that were available to all pa- patients received. dicated for nausea and emesis control. tients included additional protocol agent We used the term non-classical acupunc- prochlorperazine, lorazepam, and di- METHODS ture to describe a protocol of minimal phenhydramine, as well as metoclopra- Study Site and Patients needling near sites that are not indi- mide and droperidol. Additional medi- This study was conducted at a tertiary cated for nausea and emesis control with cations were given at the discretion of teaching hospital with a comprehen- mock stimulation. Patients were in- staff physicians not involved in this study sive cancer center. Patients were re- formed that the beneficial effect of ei- as well as at the patient’s request. cruited from oncology clinics between ther treatment is not known. Electroacupuncture. Patients in the March 1996 and December 1997 and en- Serially numbered, sealed, opaque electroacupuncture group received ad- rolled successively after written in- envelopes were used to indicate assign- junct treatment that consisted of performed consent was obtained. The as- ment. An investigator who had no direct pendicular insertion of a 36-gauge dis- sessment and intervention procedures contact with the study patients pre- posable stainless steel acupuncture were administered when the patients pared the envelopes, using a random needle (Seirin, Japan) at PC6 acupunc- were hospitalized in oncology wards for number table to generate the sequence. ture point, located between the ten- myeloablative chemotherapy. The study Patients were entered into a study log dons of palmaris longus and flexor carpii protocol was approved by the local can- before the envelopes were opened. All radialis at 2 body-inches (a body-inch or cer center scientific peer review com- envelopes were accounted for. a cun is the greatest width of a patient’s mittee and the institutional human sub- thumb at the distal phalanx) above the ject protection committee. Treatment Regimens wrist crease. (This article uses the inter- Female patients 18 to 62 years of age Chemotherapy. All patients in the 3 national nomenclature agreed on by the were eligible if they had histologically groups received the same chemo- World Health Organization in 198910). proven resected breast cancer, Karnof- therapy regimen and antiemetic drugs, The depth of insertion was 1 body- sky performance status greater than 80 following a standard protocol: On hos- inch. The needle was inserted with bi- (on 0-100 scale), life expectancy of at pital days 1, 2, and 3, all patients re- lateral rotation without introducer and least 6 months, and were appropriate ceived high doses of cyclophospha- was manipulated until achieving a “de candidates for the bone marrow trans- mide and cisplatin, and on day 4, Qi” sensation, that is, the acupunctur- plantation program. We excluded carmustine. Administration of the che- ist feels sensations from the needle-

Downloaded from www.jama.com on August 23, 2007 ELECTROACUPUNCTURE FOR CHEMOTHERAPY-INDUCED EMESIS manipulation and the patient feels sore- was given at the same time over the ensu- ing the 5-day study period. Emesis was ness, fullness, heaviness, or local area ing days for a total of 5 treatments. defined as projection of gastric con- distention. The needling technique in- Minimal Needling. For patients in the tents with resultant emesis product. Se- cluded twirling, thrusting, lifting, and ini- minimal needling group, a 36-gauge dis- vere retching without projection of gas- tial flicking. After de Qi was achieved, the posable acupuncture needle was in- tric contents was not considered as an needle was connected through a micro- serted subcutaneously with no manipu- emesis episode. The nurse’s recording of alligator clip and an electrode to a battery- lation or stimulation near the LU7 the number of daily emesis episodes was operated pulse generator connected to acupuncture point, located on the lat- used to measure emesis. Nurses were not the negative pole. A second needle was eral aspect of the radius proximal to the informed of the treatment group to inserted perpendicularly to a depth of 1.5 styloid process. Care was taken to avoid which a patient was assigned. A count body-inches at the ST36 acupuncture de Qi sensation. The needle was then of zero on a study day meant that the pa- point (located between the tibialis ante- connected through a microalligator clip tient experienced no emesis. As a sec- rior muscle and the tendon of the exten- and an electrode to a battery-operated ondary outcome, we compared the pro- sor digitorum longus pedis at 1-finger’s stimulator. A second needle was in- portion of emesis-free days across the 3 breadth lateral to the lower border of tibia serted subcutaneously, with no manipu- treatment groups. tuberosity) with a microalligator clip and lation or stimulation, near the GB34 acu- For the primary outcome, we summed an electrode connected to the positive puncture point, located in the depression the daily counts over the 5-day study pole. The needling technique included anterior and inferior to the head of the period. We also instructed patients to twirling, thrusting, lifting, initial flick- fibula, then connected to the stimula- record any adverse events that they ing, and periosteal pecking. These pro- tor. The needle insertion procedure was thought might be attributed to the study. cedures were performed bilaterally. This performed bilaterally. One acupunc- After the 5-day study period, we fol- acupunture protocol was based on prior ture consultant observed the treatment lowed up patients for an additional 9 literature, acupuncture textbooks, and procedure at the start of the study and days, for a total of 2 weeks (14 days). suggestions from consultant practition- confirmed the technique. We abstracted the number of emesis ers following a common symptomatic This minimal needling protocol was episodes from the nursing records along approach.11 designed by 2 traditional Chinese medi- with strict daily input and output re- Electrical frequency was delivered cal physicians (C.C.). In their opinion, cords as measured in milliliters (data not over 2 to10 Hz, 0.5 to 0.7 milliseconds the protocol would not harm patients reported here). We also abstracted data duration pulse width, under a variable and, at best, it could improve patients’ from medical charts on the concurrent direct current output with square wave- “lung” and “muscle-skeletal” condi- antiemetic medications administered. form balanced alternating polarity of less tions. However, they thought that the Data were collected and entered into than 26 mA for 20 minutes (maximal procedure was unlikely to prevent eme- electronic files by research assistants who voltage 15 V). Needling sites were exam- sis. The stimulator delivered the same had no knowledge of a patient’s treat- ined at the end of each treatment. Two audiovisual stimuli as in the electroacu- ment group assignment. acupuncture consultants observed the puncture technique for 20 minutes, but treatment procedure at the start of the no electrical current was passed to the Assessment of Blinding study and confirmed the technique. needles. Needling sites were examined At the end of the 5-day study period, we Two investigators (J.G. and C.C.) at the end of each treatment. asked patients who received either type administered the procedure in collabo- Treatment was administered by the of adjunct needling treatment to com- ration. One was a clinical instructor at same clinicians who administered the plete a questionnaire evaluating vari- the medical school and had 3 years of acu- electroacupuncture. The diagnostic and ous aspects of the procedures. Ques- puncture training; the other was an acu- treatment ritual was the same as that re- tions included ratings of the technical puncture clinician with 20 years of prac- ceived by the electroacupuncture group, quality of the treatment, friendliness of ticing experience. Before the needles were following the same schedule. the physician administering the treat- inserted, the clinicians evaluated patients Pharmacotherapy. Patients in the ment, and the comfort level of the pro- according to traditional pulse diagnosis pharmacotherapy alone control group cedure. In addition, we asked patients procedures. The diagnostic ritual was fol- received triple pharmacotherapy agents the treatment group to which they lowed daily, although only the standard described above and daily morning vis- thought they were assigned. electroacupuncture protocol was admin- its by physicians and nurses but re- istered. The evaluation and treatment ceived no adjunct acupuncture or mini- Statistical Analysis procedure took 30 minutes. The first mal needling therapy. Sample size and power estimation were treatment was scheduled to occur within based on 2-group comparisons of eme- the 2 hours before the initial chemo- Outcome Measures sis counts, assuming the mean count therapy infusion, usually between 7 AM The primary outcome was the total num- over the study period was 15 with Pois- and 9 AM on hospital day 1. Treatment ber of emesis episodes that occurred dur- son distribution (variance equal to the

Downloaded from www.jama.com on August 23, 2007 ELECTROACUPUNCTURE FOR CHEMOTHERAPY-INDUCED EMESIS

son models were used to examine Figure 1. Trial Profile whether the effect of treatment on the fre-

111 Eligible quency of emesis episodes was dimin- Patients ished during the follow-up period. We adjusted for the same set of baseline vari-

7 Not Randomized ables in this regression model as we did 2 Refused Randomization in the primary analysis. Third, GEE/ 4 Were Afraid of Acupuncture Needles logistic models were used to examine for 1 Had Hemothorax Before a diminished effect, treat the outcomes Randomization as either the presence or absence of emesis episodes on a given day, with and

104 Randomized without adjusting for the same set of baseline variables.

37 Received Adjunct 33 Received Adjunct Minimal 34 Received Pharmacotherapy Only RESULTS Electroacupuncture as Allocated Needling as Allocated as Allocated Patient Recruitment and Follow-up 1 Patient Missed 1 Patient Missed 1Intervention Session 1Intervention Session The recruitment and follow-up of study patients is shown in FIGURE 1.We

37 Completed Trial 33 Completed Trial 34 Completed Trial recruited 111 consecutive eligible patients, of whom 104 were randomized (2 refused randomization; 4 mean). A design with 35 patients in each the patient’s age, emesis experience with reported a fear of acupuncture needles; of the 3 groups will have 93% and 77% previous chemotherapy, alcohol use, and and 1 developed hemothorax from a power, respectively, to detect a reduc- experience with anticipatory nausea surgical catheter before randomization). tion in the mean emesis count of 25% prior to chemotherapy. The variable Thirty-seven patients were assigned to and 20% with a 2-sided test at the ␣ level selection was based on prior litera- receive electroacupuncture, 33 to mini- of .017 (adjusting for the 3 pairwise com- ture16-18 and achieving a parsimonious mal needling, and 34 to pharmaco- parisons, .05/3). Such effect sizes are model. The Wald test was used to test therapy alone. All but 1 patient received conservative estimates based on prior lit- for differences between treatment her intended chemotherapy treat- erature.12,13 groups. For all pairwise comparisons of ments (1 individual received only a par- We analyzed data according to the the 3 treatment groups, we considered tial dose of carmustine due to intoler- intention-to-treat principal, that is, based a P=.05 to indicate statistical signifi- ance) and 2 patients partly deviated on all randomized patients, as random- cance. According to the Bonferroni from their adjunct minimal needling or ized. For the primary analysis, the unit adjustment, we used PϽ.017 to adjust electroacupuncture treatment proto- of analysis was the patient. We summed for multiple comparison of 3 pairwise col (1 patient was transferred to a car- the number of emesis episodes over 5 comparisons. All P values were 2-tailed. diac care unit for cardiac toxicity and days14 and analyzed the total as a count As a secondary analysis, we com- did not receive all intended interven- variable. Because the primary outcome pared the proportion of emesis-free days tion sessions; 1 individual missed a was not normally distributed and highly across the 3 treatment groups, adjust- treatment session because of a surgical variable (ie, some patients had a substan- ing for the same set of baseline vari- procedure to replace a central catheter). tial number of emesis episodes), we ana- ables as in the primary analysis. This Two patients experienced adverse ef- lyzed the data using nonparametric tests analysis was done using a generalized es- fects as a result of the electroacupunc- and quasi-likelihood/Poisson models. timating equation (GEE)/logistic model ture or minimal needling procedure: We first compared the nonadjusted with robust variance estimation.19 Ad- 1 patient complained of an electrical outcome among 3 treatment groups ditional secondary analyses were per- shock sensation from the needle-and- using the Kruskal-Wallis test followed formed to examine the effect of treat- stimulator apparatus at the end of the by the Wilcoxon rank-sum test for pair- ment during the follow-up period; the first treatment session and the equip- wise comparisons. Then, to adjust for a outcome was the total number of eme- ment was immediately removed. The pa- common set of baseline variables pre- sis episodes that occurred during the fol- tient reported no complaints on subse- dictive of the number of emesis epi- low-up period, days 6 through 14. First, quent treatment days. One patient, who sodes and to take into account the highly we used the Kruskal-Wallis test fol- had residual peripheral neuropathy skewed distribution of the emesis data, lowed by the Wilcoxon rank-sum test to manifested as tingling and numbness we conducted multivariate analyses test for differences in the frequency of from prior chemotherapy, complained using quasi-likelihood/Poisson mod- emesis episodes among groups during of an aggravated tingling sensation fol- els.15 These baseline variables included the follow-up period. Second, GEE/ Pois- lowing each needling procedure.

Downloaded from www.jama.com on August 23, 2007 ELECTROACUPUNCTURE FOR CHEMOTHERAPY-INDUCED EMESIS

Baseline Characteristics the study period was similar across the coxon rank-sum test; PϽ.001), and the The mean age of participants was 46 3 groups (TABLE 3). During the same pe- minimal needling group had signifi- years. About two thirds of the patients riod, the use of rescue antiemetics ad- cantly fewer emesis episodes than the reported their ethnicity as white and ministered in addition to the triple agents pharmacotherapy alone group (Wil- 2 most had a college education (TABLE 1). also was comparable (Pearson ␹2=2.23; coxon rank-sum test; P=.01). Using a All patients previously had received che- P=.33; data not shown). quasi-likelihood/Poisson model to ad- motherapy, a mean of 6 courses. The vast just for potentially confounding fac- majority had experienced nausea with Outcomes tors, including age, alcohol use, emesis prior chemotherapy, and two thirds had The study outcomes are shown in experience with prior chemotherapy, experienced emesis associated with pre- TABLE 4. Total emesis episodes per per- and anticipatory nausea, the differ- vious chemotherapy. More than one son over the 5-day study period dif- ences among groups remained signifi- fourth reported anticipatory nausea as- fered among the 3 groups (Kruskal- cant. Pairwise comparisons showed that sociated with chemotherapy. Patients’ Wallis rank test; PϽ.001) (FIGURE 2). both the minimal needling and elec- baseline characteristics were well Pairwise comparisons between groups troacupuncture groups had fewer eme- matched among groups, except that pa- showed that the electroacupuncture sis episodes than the pharmacotherapy tients assigned to receive minimal nee- group had significantly fewer emesis epi- alone group (␤ coefficient for electroacu- dling had significantly more emesis with sodes than the minimal needling group puncture group=−0.36; SE, 0.068; PϽ prior chemotherapy (difference among (Wilcoxon rank-sum test; PϽ.001) or .001; and ␤ coefficient for minimal nee- groups, P=.01). the pharmacotherapy alone group (Wil- dling group=−0.12; SE, 0.063; P=.02).

Assessment of Masking Table 1. Baseline Characteristics of the Participants* Most patients reported that they did not Electroacupuncture Minimal Needling Pharmacotherapy know the group to which they were Group Group Only Group assigned (between patients in the elec- Characteristics (n = 37) (n = 33) (n = 34) troacupuncture group and those in the Age, mean (SD), y 45.5 (7.4) 43.8 (8.0) 48.0 (6.8) minimal needling group: Pearson Self-reported ethnic category, 67.6 (52.5-82.7) 63.6 (47.2-80.0) 79.4 (65.8-93.0) % white (95% CI) ␹2=4.38; P=.11) (TABLE 2). Patients’ rat- 2 Schooling, mean (SD), y 15.3 (2.5) 14.6 (3.3) 14.6 (2.3) ings of the clinician administering the Prior chemotherapy, % 100 100 100 treatment, the clinician’s friendliness, and Emesis with prior chemotherapy, 51.3 (35.2-67.4) 84.8 (72.6-97.0) 61.7 (45.4-78.0) the technical quality of the treatment they % (95% CI)† received were comparable across the Anticipatory nausea prior 27.0 (12.7-41.3) 27.3 (12.1-42.5) 23.5 (9.2-37.8) groups. The majority of the patients gave to chemotherapy, % (95% CI) ratings of excellent, and there were no Alcohol use, Ͼ5 drinks per week, 16.2 (4.3-28.1) 21.2 (7.3-35.1) 11.8 (1.0-22.6) % (95% CI) statistical differences in ratings between *CI indicates confidence interval. 2 those in the electroacupuncture group †Difference among groups, P = .01 ( ␹2 = 8.94, P = .01). and those in the minimal needling group. On a 0- to 10-point scale, with 0 being Table 2. Assessment of Masking not comfortable at all and 10 being very Electroacupuncture Minimal comfortable, patients in the minimal nee- Group Needling Group (n = 37) (n = 33) ␹2 P Value dling group had a mean rating of 9.1 for 2 When asked “Which treatment group comfort of their treatment, compared do you suspect or think you with a mean rating of 7.8 in the elec- were assigned to?” No. (%) troacupuncture procedure group (Wil- Don’t know 18 (49) 24 (73) coxon rank-sum test; P=.13). Classical* 11 (30) 6 (18) 4.38 .11 Non-classical† 8 (21) 3 (9) Other Treatments When asked to rate, % rated excellent (95% CI)‡ All patients received antiemetic medi- “The quality of acupuncture 62.2 (46.6-77.8) 48.5 (31.4-65.6) 1.32 .25 cations. Overall, when adjusted to a 24- you received” “The physician acupuncturist 75.7 (61.9-89.5) 78.8 (64.9-92.7) 0.10 .76 hour period, a patient received a mean that treated you” dose of lorazepam, 9.8 mg; diphenhy- “The friendliness of the physician 81.1 (68.5-93.7) 75.8 (61.2-90.4) 0.29 .59 dramine, 103.3 mg; and prochlorpera- acupuncturist” zine, 42.4 mg, per day, including the pro- *The term “classical (acupuncture)” was used to describe a protocol for electroacupuncture at sites classically indi- cated for control of nausea and emesis. tocol and rescue doses, during the 5-day †The term “non-classical (acupuncture)” was used to describe a protocol of minimal needling in sites that are not indi- study period. The mean amount of triple cated for control of nausea and emesis with mock stimulation. ‡CI indicates confidence interval. Rating scale: 1, very poor; 2, poor; 3, fair; 4, good; 5, very good; 6, excellent. antiemetic agents administered during

Downloaded from www.jama.com on August 23, 2007 ELECTROACUPUNCTURE FOR CHEMOTHERAPY-INDUCED EMESIS

14 were not significantly different Table 3. Concurrent Antiemetics* (Kruskal-Wallis rand test, P=.18 and Daily Mean (SD) Over Days 1-5 P=.39, respectively) We used a GEE/ Electroacupuncture Minimal Needling Pharmacotherapy Poisson regression model to test whether Group Group Only Group (n = 37) (n = 33) (n = 34) the treatment effect was diminished dur- Body-surface area, m2 (SD) 1.7 (0.16) 1.7 (0.17) 1.7 (0.15) ing the follow-up period while adjust- Lorazepam, mg (SD) 9.2 (2.07) 9.8 (1.67) 10.3 (2.16) ing for the same set of baseline variables Diphenhydramine, mg (SD) 96.9 (31.63) 105.9 (25.56) 107.2 (29.45) as in our primary analysis. Results Prochlorperazine, mg (SD) 42.8 (13.32) 40.7 (12.35) 43.7 (11.31) revealed that the difference in emesis *Dose adjusted to a 24-hour day. counts between the pharmacotherapy and the electroacupuncture groups was Table 4. Study Outcomes* diminished in the follow-up period Ͻ Electroacupuncture Minimal Needling Pharmacotherapy (P .001). All other differences (elec- Group Group Only Group troacupuncture group vs minimal nee- (n = 37) (n = 33) (n = 34) dling group and minimal needling group Study Period (Days 1-5) vs pharmacotherapy alone group) were No. of emesis episodes per person also diminished and were not statisti- Median (range) 5 (1-25) 10 (3-24) 15 (0-25) Mean (95% CI)† 6.29 (4.20-7.02) 10.73 (7.38-11.90) 13.41 (9.55-15.05) cally significant. Similar results were Percent emesis-free days found using the GEE-binary models to Mean (95% CI) 55 (47-63) 29 (20-37) 20 (11-29) treat the outcome as either the presence Follow-up Period (Days 6-14) or absence of an episode on a given day. No. of emesis episodes per person In summary, during the follow-up pe- Median (range) 4 (0-32) 7 (0-30) 8 (0-22) riod days 6 to 14, when groups were no Mean (95% CI)† 6.89 (3.65-7.34) 8.60 (4.84-9.42) 8.56 (5.29-9.48) longer receiving adjunctive therapy, Percent emesis-free days, 60 (52-68) 53 (45-62) 52 (44-62) there were no significant differences mean (95% CI) among groups in the number of eme- *CI indicates confidence interval. †Constructed on square root scale and back transformed to the original scale. sis episodes or proportion of emesis- free days. puncture group minus minimal nee- Figure 2. Distribution of the Total Emesis COMMENT Episodes Per Person During the 5-Day Study dling group=−0.24; SE, 0.073; PϽ.001). Period by Treatment Groups As a secondary analysis, patients in the The results of this study suggest that the electroacupuncture group had a greater addition of daily electroacupuncture P < .001 P = .01 proportion of emesis-free days than pa- treatment to this antiemetic regimen was 25 tients in either the minimal needling superior to the pharmacotherapy therapy group or the pharmacotherapy group alone or minimal needling in prevent- 20 (Kruskal-Wallis rank test, PϽ.001). A ing chemotherapy–induced emesis. 15 GEE/logistic model adjusting for the There also was a trend indicating that same baseline variables as in the pri- the minimal needling procedure itself 10 mary analysis showed that the propor- was more effective in reducing emesis 5 tion of emesis-free days was signifi- episodes than pharmacotherapy alone. Total Emesis Episodes Total cantly greater in the electroacupuncture The observed differences between groups 0 Electroacupuncture Minimal Pharmacotherapy group compared with either the mini- diminished in the follow-up period, (n = 37) Needling Only (n = 33) (n = 34) mal needling group (PϽ.001) or the which further supported an antiemetic pharmacotherapy alone group (PϽ.001). effect of electroacupuncture. The lines inside the box represent the median values. The proportion of emesis-free days was Our study has several strengths and The lower and upper borders of the boxes represent the 25th and 75th percentiles. The interquartile range not significantly lower in the minimal limitations. A principal strength is our is the height of the box. The whisker lines extend from needling group as compared with the use of a minimal needling intervention the box borders to data points that are less than or equal to 1.5 interquartile ranges. Circles outside the whisker pharmacotherapy alone group (P=.18). to assess the potential nonspecific ef- lines represent extreme values. P values are based on We also used 3 sets of analyses to fects of needling and attention and care pairwise tests using the Wilcoxon rank-sum test. address the issue of whether the treat- for the patient. While we could not mask ment effect from the study period dimin- the acupuncturists to the interventions Furthermore, the electroacupuncture ished during the follow-up period. Over- they were delivering, we were able to group had fewer emesis episodes than all, comparing all 3 groups, the sums of achieve adequate masking of the pa- the minimal needling group (the differ- emesis episodes as well as the propor- tients. This strengthens the internal va- ence in the ␤ coefficients for electroacu- tion of emesis-free days during days 6 to lidity of our study and increases the like-

Downloaded from www.jama.com on August 23, 2007 ELECTROACUPUNCTURE FOR CHEMOTHERAPY-INDUCED EMESIS lihood that the effect we observed was therapy setting, emetic responses are sea and Vomiting. New York, NY: The Parthenon Pub- lishing Group; 1993:93-109. due to a specific effect of electroacupunc- aggravated by the multiple agents, their 10. Jenkins M. A new standard international acupuncture. A second strength is our use of an converted products, altered metabo- ture nomenclature. Acupuncture in Medicine. 1980; 7:21-25. electroacupuncture protocol that was lism, and products of cell damage. The 11. O’Connor J, Bensky D, trans-eds. Acupuncture: A standardized, reproducible, and in agree- complex, multifactorial and severe nature Comprehensive Text: Shanghai College of Tradi- ment with the practices of our consult- of such chemotherapy–induced emesis tional Medicine. Chicago, Ill: Eastland Press; 1981: 557. 12. Dundee JW, Ghaly RG, Fitzpatrick KT, Abram WP, ant practitioners, making it more likely suggests that no single antiemetic agent Lynch GA. Acupuncture prophylaxis of cancer che- that others can attempt to replicate our targeting a particular mechanism can be motherapy-induced sickness. J R Soc Med. 1989;82: 268-271. results. Such attempts at replication are expected to control vomiting com- 13. Aglietti L, Roila F, Tonato M, et al. A pilot study of needed as we used only 2 acupunctur- pletely. Electroacupuncture (repeated metoclopramide, dexamethasone, diphenhydramine and acupuncture in women treated with cisplatin. Cancer ists working at a single hospital. sensory stimulation) has been thought Chemother Pharmacol. 1990;26:239-240. The homogeneity of the patient popu- to modulate serotonin, substance P, and 14. Matthews J, Altman D, Campbell M, Royston P. Analysis of serial measurements in medical research. lation and their receipt of a standard pro- endogenous opiates along various path- BMJ. 1990;300:230-235. 24-27 tocol for chemotherapy and supportive ways in the central nervous system. 15. McCullagh P, Nelder JA. Generalized Linear Mod- care increased the precision with which We speculate that some of the effects we els. London, England: Chapman & Hall; 1989. 16. Gralla RJ, Clark RA, Kris MG, Tyson LB. Method- we could measure the treatment effects observed may be manifested through the ology in antiemetic trials. Eur J Cancer. 1991;27(suppl but limits the generalizability of our find- serotonin- and substance P–mediated 1):S5-S8. 17. Morrow G. Behavioural factors influencing the de- ings to other patient populations or those components of the emetic reflex, as well velopment and expression of chemotherapy induced receiving other adjunct therapies. The an- as through the opiate µ receptor via its side effects. Br J Cancer Suppl. 1992;66(suppl 19):S54- 28-30 S60; discussion S60-S63. tiemetic pharmacotherapy used in our antiemetic actions. Future neuro- 18. Tonato M, Roila F, DelFavero A, Ballatori E. Meth- study protocol does not include corti- physiological and neurochemical inves- odology of trials with antiemetics. Support Care Can- cer. 1996;4:281-286. costeroids or a serotonin antagonist, such tigations may help us to further under- 19. Liang K, Zeger S. Longitudinal data-analysis using 20 as ondansetron. Use of these agents has stand the complexity of emesis and to generalized linear models. Biometrika. 1986;73:13-22. been shown to be superior to the agents broaden the current approach to the spec- 20. Gralla RJ, Osoba D, Kris MG, et al, for the Ameri- can Society of Clinical Oncology. Recommendations for used in this study for controlling eme- trum of antiemetic care. the use of antiemetics. J Clin Oncol. 1999;17:2971- sis. The effect of electroacupuncture as 2994. Funding/Support: This work was supported in part 21. Lao L, Berman B, Wong R, Wynn R. Effect of elec- an adjunct to other antiemetic regi- by grant PE19001-09, Health Resources and Ser- troacupuncture as an adjuvant for treating cyclophos- mens, including serotonin antagonists vices Administration, and grant 41897 to Dr Shen from phamide-induced emesis in ferrets. Proceedings De- the American Cancer Society, California Division Inc. partment of Defense Breast Cancer Research Program and corticosteriods, is unknown. Acknowledgment: We thank Ellen Karpf and Linda Meeting, Vol 3. Washington, DC: US Government Print- It is important to note that minimal Norton for patient recruitment and William Harlan, MD, ing Office; 1997:861-862. needling led to a reduction in the fre- for reviewing the manuscript and providing com- 22. Grunberg SM, Hesketh PJ. Control of chemo- ments. Dr Shen thanks the National Center for Comple- therapy-induced emesis. N Engl J Med. 1993;329: quency of emesis episodes. Attention and mentary and Alternative Medicine for their intramural 1790-1796. the clinician-patient interaction are pos- support. 23. Navari RM, Reinhardt RR, Gralla RJ, et al, for the L-754,030 Antiemetic Trials Group. Reduction of sible explanations for the beneficial ef- cisplatin-induced emesis by a selective neurokinin-1- fect in this setting. This finding sup- REFERENCES receptor antagonist. N Engl J Med. 1999;340:190-195. ports a role for behavioral interventions 1. Cagnoni PJ, Matthes S, Day TC, Bearman SI, Shpall 24. Han JS, Xie GX, Zhou ZF, Folkesson R, Terenius L. EJ, Jones RB. Modification of the pharmacokinetics of Acupuncture mechanisms in rabbits studied with mi- concomitant to pharmacological man- high-dose cyclophosphamide and cisplatin by anti- croinjection of antibodies against beta-endorphin, en- agement and also suggests that, in fu- emetics. Bone Marrow Transplant. 1999;24:1-4. kephalin and substance P. Neuropharmacology. 1984; 2. Gilbert CJ, Petros WP, Vredenburgh J, et al. Phar- 23:1-5. ture studies evaluating the efficacy of macokinetic interaction between ondansetron and cy- 25. Vacca-Galloway LL, Naftchi NE, Arakawa K, Guan acupuncture, a convincing control and clophosphamide during high-dose chemotherapy for XM, Ai MK. Alterations of immunoreactive substance breast cancer. Cancer Chemother Pharmacol. 1998; P and enkephalins in rat spinal cord after electroacu- successful masking are critical. 42:497-503. puncture. Peptides. 1985;6:177-188. Our study showed that adding a daily 3. Frigerio C, Buchwalder PA, Spertini F. Ondanse- 26. Cheng RS. Neurophysiology of electroacupunc- tron: reasons to be restrictive. Lancet. 1996;347:1484- ture analgesia. In: Pomeranz B, Stux G, eds. Scientific electroacupuncture procedure to phar- 1485. Basis of Acupuncture. New York, NY: Springer- macotherapy was more effective than 4. NIH Consensus Conference: acupuncture. JAMA. Verlag; 1989:119-135. pharmacotherapy alone in preventing 1998;280:1518-1524. 27. Bucinskaite V, Lundeberg T, Stenfors C, Ekblom A, 5. Dundee JW, Ghaly RG, Fitzpatrick KT, Lynch GA, Dahlin L, Theodorsson E. Effects of electro- chemotherapy–induced emesis. Similar Abram WP. Acupuncture to prevent cisplatin- acupuncture and physical exercise on regional concen- results recently have been observed in an associated vomiting. Lancet. 1987;1:1083. trations of neuropeptides in rat brain. Brain Res. 1994; 666:128-132. 21 6. Vickers AJ. Can acupuncture have specific effects animal model. Are there biologically on health? J R Soc Med. 1996;89:303-311. 28. Harris AL. Cytotoxic-therapy-induced vomiting is plausible explanations for the observed 7. Lee A, Done M. The use of nonpharmacologic tech- mediated via enkephalin pathways. Lancet. 1982;1: niques to prevent postoperative nausea and vomiting: 714-716. effect? Pharmacological therapies to man- a meta-analysis. Anesth Analg. 1999;88:1362-1369. 29. Kobrinsky NL. Regulation of nausea and vomiting age chemotherapy–induced emesis have 8. Leslie R, Reynolds D. Functional anatomy of the in cancer chemotherapy: a review with emphasis on opi- emetic circuitry in the brainstem. In: Bianchi AL, Gre- ate mediators. Am J Pediatr Hematol Oncol. 1988;10: been directed at the neurotransmitter lot L, Miller AD, King GL, eds. Mechanisms and Con- 209-213. receptors in the brain regions receptive trol of Emesis. London, England: INSERM/ John Libbey 30. Rudd JA, Cheng CH, Naylor RJ, Ngan MP, Wai MK. 22,23 Eurotext Ltd; 1992:19-27. Modulation of emesis by fentanyl and opioid receptor to emetic stimuli. In a multiple-day, 9. Epstein BS. Preventing postoperative nausea and antagonists in Suncus murinus (house musk shrew). Eur multiple-drug combination chemo- vomiting. In: Sleisenger MH, ed. The Handbook of Nau- J Pharmacol. 1999;374:77-84.

Downloaded from www.jama.com on August 23, 2007