A Randomized, Placebo-Controlled, Double-Blind Study of Minocycline for Reducing the Symptom Burden Experienced by Patients with Advanced Pancreatic Cancer

Abstract

Context:

Although it’s well known that patients with advanced pancreatic cancer (PC) experience significant symptom burden, few strategies for effective symptom intervention are available for them.

Objective:

To investigate the efficacy of minocycline, an anti-inflammatory agent, for symptom reduction in advanced PC patients.

Methods:

We conducted a phase II, randomized, placebo-controlled trial to obtain preliminary estimates of the effects on symptom reduction with 100 mg minocycline or placebo given twice a day. Eligible patients had diagnosed advanced PC and were scheduled for standard chemotherapy. Patient-reported symptoms were measured weekly during the 8-week trial using the MD Anderson Symptom Inventory module in patients with gastrointestinal cancer (MDASI-GI). The primary outcome measure was the area under the curve (AUC) values of the 5 most-severe symptoms in the 2 arms.

Results:

Of the 44 patients recruited, 31 (71%) were evaluable for the primary efficacy analysis, with 18 received minocycline and 13 placebo. Fatigue, pain, disturbed sleep, lack of appetite, and drowsiness were the most-severe symptoms reported by both groups. No significant differences in AUC values over time between the study arms were found for the composite MDASI score or single-item scores of the 5 most-severe MDASI items. No treatment-related deaths were reported, and no grade 3–4 toxicities were observed.

Conclusion:

Minocycline is safe for use in patients receiving treatment for PC. There is no observed symptom reduction with minocycline on the major symptom burden associated with advanced PC compared with placebo. Attrition due to rapid disease progression impacted the study significantly.

INTRODUCTION

Pancreatic cancer (PC) is a well-known, highly symptomatic, and rapidly lethal illness (1, 2). Most patients are diagnosed with advanced disease and experience accelerated tumor progression and high symptom burden within a few months, regardless of treatment (3–5). Given that progression-free survival is often less than 2 months, most patients with PC experience a rapid and merciless course of intractable pain, fatigue, poor appetite, distress, and death within the first 12 months of diagnosis (3, 6). The lack of effective approaches to symptom control produces enormous frustration for medical professionals and a great burden to patients and families who are dealing with advanced PC (7, 8).

The clinical observations in patients with advanced PC, either driven by rapid disease progression or aggressive cancer therapy, often appear as a cluster of significant symptoms burden (1, 4, 6, 9, 10). Furthermore, animal models of sickness behavior confirmand the link between the tumor progression and the symptom burden in animals with tumors; poor survival and accelerated tumor growth were observed after the animals experienced pain or stress (11, 12). Thus, symptom management is one of the goals of personalized clinical cancer care that effectively guide treatment decision for both unresectable and metastatic PC (13, 14).

There is a growing research effort to understand symptoms’ underlying biological mechanisms and effectively control the symptom burden in patients with advanced cancer. Inflammation is a well-known, key mechanism of disease progression that is now being investigated as a mechanism underlying the development of multiple severe symptoms (15–20).

We designed the current study on the basis of growing evidence from preclinical and clinical studies of the role of anti-inflammation in PC, and from the potential clinical benefit of using minocycline for disease and symptom control in patients with cancer (21, 22) (23, 24) and other illnesses (25–28). Minocycline showed ability to modulate inflammatory cascades, and inhibit hypoxia-induced changes and cytokines release, in preclinical studies (29–32). In term of therapeutic benefits, clinical studies proof that minocycline is safe and its use has positive trend toward reducing cancer-related symptoms burden (neuropathy pain, dermatitis, diarrhea and feeding tube dependency ) in head and neck, lung and bresat cancer patients (21, 22) (23, 24).

In this study, we hypothesized that the anti-inflammatory effects of minocycline—a safe, inexpensive, readily available agent—would be effective in reducing symptom expression in patients with advanced PC. To test this hypothesis, we conducted a proof-of-concept phase II study.

METHODS

Study design and Participants

We conducted a randomized, 2-arm, double-blind, placebo-controlled trial. The study was approved by The University of Texas MD Anderson Cancer Center Institutional Review Board and is registered at Clinicaltrials.gov (NCT01693523).

Eligible patients had newly diagnosed PC; qualified for gemcitabine-based or FOLFIRINOX-based first-line chemotherapy; aged >18 years old; able to speak English or Spanish; able to review, understand; able to provide written consent before starting therapy; had an Eastern Cooperative Oncology Group score of 0 to 2; with adequate renal and liver function, and agreed to stop using nutritional supplements. Exclusion criteria include: patients who were enrolled in other symptom management clinical trials; and patients who, in the judgment of the investigator, may be unable to participate in the required study procedures, patients who were taking medication or had conditions that potentially preclude use of minocycline, as determined by the treating physician; patients who had bile duct obstruction or cholelithiasis; patients with hypersensitivity to any tetracyclines; patients who were pregnant; patients who were under treatment of warfarin with INR > 1.5.

Patient Enrollment and Registration

Patients were screened for eligibility and recruited for enrollment in the outpatient GI Medical Oncology Clinic. The opportunity to participate in the minocycline trial was presented to eligible patients. Those who opt not to enroll in the minocycline study, along with those who do not qualify for the minocycline study, were approached about participating in the observational arm. Research staff maintained a log of all patients screened, and the reasons that patients do not enter the study were documented. Enrolled patients were registered into the Clinical Oncology Research System (CORe), the institutional patient data management system. All participants gave written informed consent. Eligible patients were followed in the GI Medical Oncology Clinic during routine clinic visits.

Randomization, blinding and Intervention

Prior to accruing the first patient, a randomization list for the entire sample was generated by a qualified biostatistician, stating into which group a patient will be randomized. This list containing the accrual number and treatment group information was given to Investigational Pharmacy. A sealed backup list was kept by the assigned data analyst at the Department of Symptom Research. This list is only allowed to be opened in case unblinding is needed (In the event of severe AE or an emergency situation that is likely due to the symptom trial agents as determined by the treating physician or PI). Patients were randomly assigned to receive minocycline (200 mg/day) or placebo for 4 to 6 weeks, beginning on the first day of chemotherapy. Study staff contacted patients every two weeks to check for adverse events during routinely scheduled clinic visits or via telephone calls.

Patient-Reported Outcomes Tool

We monitored multiple symptoms on weekly basis for 8 weeks using the validated gastrointestinal cancer module of the MD Anderson Symptom Inventory (MDASI-GI) (33) to simultaneously document symptom prevalence, severity, and interference with daily functioning and life activities. The MDASI-GI contains 24 items that describe the extent to which symptoms have interfered with various aspects of the patient’s life during the past 24 hours, with 0 being “no interference” and 10 being “interfered completely.” Patients in the placebo arm were under the same assessment schedule as patients in the minocycline trial.

Statistical Analysis

Per protocol, the primary outcome measure was area under the curve (AUC) values (34) for the 5 most-severe symptoms at 4 weeks, although the PRO assessment last for 8 weeks. The AUC was computed for each of the 5 MDASI-GI symptoms using the trapezoidal rule. The average AUC value for each patient was computed by taking the mean of that patient’s 5 single-symptom AUC values. Summary statistics were then produced for each single-symptom AUC value as well as for the average AUC value. Higher AUC values indicate higher symptom burden. T-tests were used to compare the AUCs during 4 weeks for the 5 most-severe symptoms between treatment groups. The longitudinal symptom data were graphed by means and 95% confidence intervals over the 8 weeks of investigation.

We estimated that with evaluable patients per treatment arm for 4 weeks, the minimum effect size of difference between the 2 treatments would be 1 standard deviation (SD), with 80% power and a 2-tailed 5% significance test for the primary outcome. An evaluable patient for the primary outcome was one who completed at least 2 MDASI-GI assessments.

RESULTS

Study Sample

The expected sample size was 40 evaluable patients. Between Feb 2013 and Oct 2015, we approached 63 patients for screening the eligibility; 44 were enrolled and randomly assigned to receive minocycline (n = 23) or placebo (n = 21), Supplementary Fig 1.

Summary of Reasons for Exclusion from Primary Efficacy Analysis & Attrition Rate:

Of the 44 participants, 31 (71%) were evaluable for the primary efficacy analysis and completed expected MDASI-GI assessments, Supplementary Fig 1. Table 1 includes patient and disease characteristic.

Table 1.

Patient and Disease Characteristics (N=31)

Patient characteristics, mean (SD)	Placebo (n=13)	Minocycline (n=18)	P-value (Chi-sq or Fisher’s exact)
Age, years	59.92 (11.82)	60.44 (7.55)	0.8818
Education grade	15.62 (2.36)	14.78 (2.10)	0.3073
Patient characteristics, No. (%)
Sex			0.3473
Women	5 (38.46)	10 (55.56)
Men	8 (61.54)	8 (44.44)
Race			0.6989
White	8 (61.54)	11 (61.11)
Hispanic	3 (23.08)	4 (22.22)
Others	2 (15.38)	3 (16.67)
Employment			0.6525
Employed	4 (30.77)	9 (50.00)
Retired	5 (38.46)	5 (27.78)
Others	4 (30.77)	4 (22.22)
Marital Status			0.1325
Married	9 (69.23)	11 (61.11)
Divorced	3 (23.08)	1 (5.56)
Others	1 (7.69)	6 (33.33)
ECOG
Good (0–1)	13 (100.00)	18 (100.00)
Poor (2–3)	0 (0.00)	0 (0.00)
Charlson Comorbidity Index			0.9248
0	7 (53.85)	10 (55.56)
1+	6 (46.15)	8 (44.44)
Stage			0.6061
I	3 (23.08)	1 (5.56)
II	1 (7.69)	2 (11.11)
III	1 (7.69)	3 (16.67)
IV	8 (61.54)	12 (66.67)

Abbreviations: ECOG, Eastern Cooperative Oncology Group; SD, standard deviation.

Reasons for exclusion

Minocycline group:

2 patients stopped receiving chemotherapy, 1 patient decided to stop taking the study medication, 1 patient was removed due to the disease progression, and 1 patient was removed due to lack of compliance with the study medication.

Placebo group:

1 patient never took the study medicine or completed the MDASI-GI, 1 patient requested to stop chemotherapy, 2 patients requested to be removed from the study, 1 patient was removed due to lack of compliance with the study medication, 1 patient had the study medication discontinued due to increased alanine aminotransferase level, and 1 patient was removed due to disease progression.

The Major Symptom Burden of Patients in the Trial

Fatigue, lack of appetite, drowsiness, pain, and disturbed sleep were the most-severe symptoms at baseline as well as over time for both groups. None of these symptoms differed in severity at baseline (pretreatment) between the 2 treatment groups. Over time, mixed modeling showed that the top 5 symptom’s severity did not change significantly for either group, and no significant difference was seen between the pre- and posttreatment mean score of symptom PROs for either group.

Efficacy of Minocycline Intervention on Symptom Burden

There were no significant differences between study arms in mean AUC values for any of the 5 individual MDASI-GI symptoms or for the average of the 5 individual symptom scores. Table 2 presents the AUC values for the evaluable patients in each group over 4-weeks trial.

Table 2.

AUC Results for Patients Randomly Assigned To Receive Minocycline or Placebo over 4-Weeks of Treatment Trial

	Placebo (N = 13)					Minocycline (N = 18)					p
	Mean	SD	Median	Min	Max	Mean	SD	Median	Min	Max
Symptom outcome on MDASI-GI
Mean, 5 most-severe symptoms	62.43	36.53	45.38	30.92	153.42	86.27	39.63	83.42	10.50	172.67	0.099
Pain	70.81	48.08	84.00	0.00	171.50	78.56	51.46	73.50	7.00	203.00	0.674
Fatigue	82.12	54.40	59.50	21.00	192.50	120.36	64.56	126.00	0.00	217.00	0.093
Sleep	56.54	46.95	35.00	0.00	157.50	81.28	52.38	91.00	0.00	168.00	0.186
Lack of appetite	71.62	60.14	52.50	0.00	178.50	88.67	79.77	66.50	0.00	234.50	0.522
Drowsiness	60.58	54.00	38.50	0.00	192.50	83.42	61.52	75.25	0.00	217.00	0.293
Interference to daily functioning on MDASI-GI
Mean, total interference	65.25	38.52	63.00	19.25	157.50	78.58	47.65	65.63	0.00	155.05	0.413
WAW	69.64	46.84	49.00	12.83	169.17	91.71	56.93	89.25	0.00	198.33	0.262
Walking	58.15	54.68	28.00	0.00	192.50	73.11	59.18	63.00	0.00	192.50	0.480
Activity	83.73	42.35	77.00	17.50	161.00	106.36	60.22	101.50	0.00	217.00	0.255
Work	67.04	54.44	52.50	0.00	157.50	95.67	59.72	84.00	0.00	206.50	0.183
REM	60.85	35.30	51.33	19.83	145.83	65.43	49.52	61.83	0.00	149.33	0.778
Relations with others	35.27	36.72	31.50	0.00	108.50	43.75	48.61	33.25	0.00	150.50	0.601
Enjoyment of life	84.00	45.90	80.50	24.50	171.50	96.06	68.00	92.75	0.00	252.00	0.584
Mood	61.65	39.97	52.50	7.00	157.50	55.03	52.39	42.00	0.00	157.50	0.705

Abbreviations: AUC, area under the curve; REM, a composite score of affective interference (interference with relations with others, enjoyment of life, and mood); WAW, a composite score of physical interference (interference with walking, daily activities, and working).

In addition, there was no significant difference in daily functioning capacity (measured by MDASI-GI interference scores) between the minocycline and placebo groups.

Mixed modeling over 7 weeks revealed no difference in the mean score of single-item quality of life between the 2 arms.

Adverse Events

A total of 25 adverse events were reported by patients while enrolled in the study: 3 possibly-attributable grade 1 event (in the minocycline group, two patients reported diarrhea and one patient reported nausea), 5 unlikely-attributable grade 1 (4 in minocycline and 1 in placebo group respectively) 8 unlikely-attributable grade 2 event (in placebo group), and 7 unlikely grade 3 events (4 in the minocycline group and 3 in placebo group), in addition to one unrelated grade 2 and one unrelated 3 events in the placebo arm, Table 3. A review of the cases of deceased patients with the treating physicians found no treatment-related deaths.

Table 3.

Reported Toxicities per Trail Groups

CTC-AE Grade	AE Attribution per Trial Groups								Total
	Probable		Possible		Unlikely		Unrelated
	Minocycline arm	Placebo arm	Minocycline arm	Placebo arm	Minocycline arm	Placebo arm	Minocycline arm	Placebo arm
1	0	0	3	0	4	1	0	0	8
2	0	0	0	0	0	8	0	1	9
3	0	0	0	0	4	3	0	1	8
4	0	0	0	0	0	0	0	0	0
Total	0	0	3	0	8	12	0	2	25

Abbreviations: CTC-AE, Common Toxicity Criteria-Adverse Events

DISCUSSION

Using minocycline to target the underlying inflammation pathways might challenge existing insufficient methods for controlling severe sickness symptoms (such as fatigue, pain, and drowsiness). However, the current randomized study did not provide evidence of a positive benefit of minocycline on symptom reduction in patients with advanced PC at week 4 during standard chemotherapy. The consistent lack of benefit of using minocycline on reducing any of the most-severe symptoms supports our conclusion of a negative result for this approach in patients with advanced PC. However, several discussion points stem from this negative trial.

First, it is well known that patients with advanced PC experience refractory disease and severe symptom burden. With the use of FOLFIRINOX as the standard oncology care at the time of this study (4, 35, 36), the driving forces of the overall symptom burden are likely both disease burden and chemotherapy side effects. Our study used an anti-inflammatory agent in an attempt to manage a cluster of symptoms, the worst of which was fatigue in this patient sample. There is still no effective intervention for fatigue in current oncology care, especially for patients with such advanced disease. While minocycline may be too weak as a candidate agent for further study in such patients, especially with the used dose of 200 mg/day, we cannot rule out that more potent agents could be considered for this anti-inflammatory mechanism–driven symptom intervention strategy.

Second, in this valuable study, we documented a complete and longitudinal symptom profile from the perspective of patients with advanced PC, using the validated MDASI-GI module, which includes both symptom severity and interference. This comprehensive information of major symptom burden over time during therapy could be used for end-of-life discussions between cancer care providers and patients, although the study sample comprised patients who were eligible for first-line chemotherapy as standard care and had good performance status. Regardless of the tumor response, the PROs provide the major symptom burden from the patients’ perspectives and involve patients in decisions about treatment for their advanced PC (37). Our results showed that the most-severe symptoms at baseline remained the worst during therapy; these observations support current American Society of Clinical Oncology and National Comprehensive Cancer Network clinical practice guidelines that emphasize that the importance of full assessment of symptom burden as early as possible in patients with locally advanced and metastatic PC, preferably during the first visit (38) (13, 14, 39, 40).

Third, this trial demonstrated the acceptable safety of minocycline for patients with advanced PC. This finding aligns with the results of our research effort and others which provide evidence that using minocycline to reduce cancer-related symptoms burden is safe and feasible (21, 22) (23, 24). However, our data showed that the minocycline group had a trend of higher symptom burden, measured by the mean AUC values of the 5 most-severe symptoms, compared with the placebo group. Other studies showed either positive trend in term of potential symptom reduction after using minocycline in cancer cohorts undergoing radiation or concurrent chemoradiation therapy, or no effect on symptom reduction in patients undergoing chemotherapy (21, 22) (23, 24). Yet, the using of minocycline to reduce symptom burden in cancer patients is still evolving and require more solid evidence from large randomized studies. The effort to study symptoms reduction in patients with advanced cancers is limited by high attrition rate and poor tolerability due to the cumulative treatment effects that those patients with advanced cancers usually experience. Likewise, in our study, the attrition rate was high and the patients experienced high symptom burden which is difficult to control. Also, the existing data that show positive trend toward reducing symptoms burden in cancer patients after using minocycline come from single arm Phase II or Phase III with small sample size and most of those studies were unpowered to detect statistically significance reduction effect in the cancer-related symptoms burden after minocycline. Yet, this positive trend in some of those studies may result from including patients with early cancer stages and excluding patients who were heavily treated (21–23).

Finally, there were several limitations and difficulties we encountered in conducting this symptom-control trial in this tertiary cancer center among patients with advanced cancer. We experienced slow enrollment due to a small number of potentially eligible patients; reasons for this included a preference by some patients to be treated with standard first-line therapy in their local clinic after receiving the oncology consultation, ineligibility for the this trial due to existing conditions from the advanced disease, and participation in cutting-edge clinical trials for cancer control that disallowed some patients from participating in this symptom intervention study. Therefore, we did not reach the desired sample size during the funding period and study did not include enough patients for the needed statistical power to show the difference between the two study arms. Also, the attrition rate was 28%, which was mostly due to rapid disease progression. Last, our statistical approach is called “per protocol”, which means that only patients who completed protocol requirements are analyzed. It has been criticized because it does not reflect true-life scenarios. However, given high attrition rate for these advanced PC, we decided to replace patients with the same treatment if that patient was considered inevaluable. In contrast to intent-to-treat analysis, which requires including patients even if they have only baseline symptom assessment, replacing patients with least than two symptom assessments was deemed a more reasonable approach compared to imputing symptom data when only baseline symptom data is available.

In conclusion, lack of positive trend in term of symptom reduction after using of minocycline from this randomized trial in patients with advanced PC suggests that no more studies are needed to test the usefulness of minocycline at the tested dose for symptom reduction in patients with advanced PC.