Vol. 2, 659-668, April 1996 Clinical Cancer Research 659
Phase I Clinical Trial of the Flavonoid Quercetin: Pharmacokinetics and Evidence for in Vivo Tyrosine Kinase Inhibition1
David R. Ferry,2 Anna Smith, Joy Malkhandi, tyrosine kinase activity, and evidence of antitumor activity David W. Fyfe, Philippa G. deTakats, was seen. David Anderson, Jim Baker, and David J. Kerr INTRODUCTION Cancer Research Campaign Institute for Cancer Studies. University of Birmingham, Queen Elizabeth Hospital, Edgbaston, Birmingham Bl5 Flavonoids are found as either simple or complex glyco- 2TH, United Kingdom sides in all aerial plants (I ). Humans have been estimated to consume approximately 1 g flavonoidsiday, the most common flavonoid glycones found in the diet being quercitrin, turin, and ABSTRACT robinin (2, 3). Quercitrin and rutin are hydrolyzed to quercetin We have performed a Phase I clinical trial with the and robinin to kempferol by the 3-glucosidase activity of obli- naturally occurring flavonoid quercetin (3,3’,4’,5,7-penta- gate anaerobes in the gastrointestinal tract (4). A recent detailed hydroxyflavone). Quercetin has antiproliferative activity in estimation of free flavonoid consumption in Western diets found vitro and is known to inhibit signal transduction targets the richest sources of quercetin to be onions (347 mg/kg), apples including tyrosine kinases, protein kinase C, and phosphati- (36 mg/kg), and red wine ( 1 1 mg/kg; Ref. 5). Since the realiza- dyl inositol-3 kinase. Quercetin was administered by short tion that many folk medicines still in use contain flavones (2), i.v. infusion at escalating doses initially at 3-week intervals. interest in this class of compounds has intensified. The first dose level was 60 mg/rn2; at the 10th dose level of Quercetin (Fig. 1) has a wide range of biological activities 1700 mg/m2, dose-limiting nephrotoxicity was encountered, including inhibition of the Na K ATPase (6), protein kinase C but no myelosuppression. At the preceding dose level of 1400 (7), tyrosine kinases (8), HIV reverse transcriptase (with a Km of mg/rn2, five patients were treated at 3-week intervals, and 0.08 p.M; Ref. 9), Ca2tATPase of sarcoplasmic reticulum (10), another eight patients were treated on a once-weekly sched- and pp60 kinase. Indeed, quercetin was probably the first ule; overall, 2 of 10 evaluable patients had renal toxicity, 1 at tyrosine kinase inhibitor to be described ( I 1), and is also re- grade 2 and 1 at grade 4. We therefore treated other patients ported to cause cell cycle arrest of human leukemic T cells (12) at 945 mg/m2 (eight at 3-week intervals and six at weekly and gastric cancer cells (13) in late G . More recently, quercetin intervals); 3 of 14 patients had clinically significant renal has been reported to induce apoptosis through a pathway in- toxicity, 2 patients with grade 2 and 1 patient with grade 3. volving heat shock proteins (14). Quercetin can down-regulate Patients treated on the weekly schedule did not have cumu- mutant p53 levels (15), and as mutant p53 can block apoptosis; lative renal impairment but did have a fall in the glomerular this is another mechanism through which quercetin could facil- filtration rate of 19 ± 8% in the 24 h after drug adminis- itate cell death. Quercetin is also a potent inhibitor of phosphati- tration. We recommend 1400 mg/m2 as the bolus dose, dyl-3 kinase ( 16) and l-phosphatidylinosotol-4 kinase (17), which may be given either in 3-week or weekly intervals, for important enzymes in proliferation signaling pathways ( I 8). Phase II trials. Quercetin has antiproliferative activity in vitro against Quercetin pharmacokinetics were described by a first- ovarian (19), breast (19), and stomach (20) cancer cell lines. In order two-compartment model with a median t#{189}a of 6 mm vivo synergy of quercetin with cisplatin against Walker lung and median t#{189}(1 of 43 mm. The median estimated clearance cancer xenografts in nude mice has been described (2 1 ). Fur- was 0.28 liter/minim2, and median volume of distribution at thermore, quercetin has antiproliferative activity against human steady state was 3.7 liter/rn2. In 9 of 11 patients, lymphocyte ovarian cancer primary cultures and can potentiate the action of protein tyrosine phosphorylation was inhibited following cisplatin ex vito (22). In primary cultures of human acute administration of quercetin at 1 h, which persisted to 16 h. myeloid leukemia, quercetin has been demonstrated to potenti- In one patient with ovarian cancer refractory to cisplatin, ate the cytotoxic action of 143-o-arabinofuranosylcytosine (23). following two courses of quercetin (420 mg/rn2), the CA 125 These activities of quercetin make it a promising candidate for had fallen from 295 to 55 units/mI, and in another patient Phase I evaluation in cancer patients. with hepatoma, the serum a-fetoprotein fell. In conclusion, Quercetin was administered in an ethanol vehicle to six quercetin can be safely administered by i.v. bolus at a dose normal human volunteers in 1975 at a fixed dose of 100 mg injection. The plasma levels achieved inhibited lymphocyte without side effects (24). We planned a dose escalation Phase I trial using 60 mg/m2 as the starting dose. We have developed a novel HPLC3 method of detecting quercetin in plasma (25), and
Received 8/3/95; revised 1 1/14/95: accepted I 2/1 1/95. I Supported by the Cancer Research Campaign. D. W. F. is a Cancer Research Campaign funded research fellow. P. G. d. T. is a senior 3 The abbreviations used are: HPLC, high-performance liquid chroma- Cancer Research Campaign clinical research fellow. tography; AUC, area under the curve; AFP, a-fetoprotein: lC ), 50% 2 To whom requests for reprints should be addressed. inhibitory concentration: NO, nitrous oxide.
Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 1996 American Association for Cancer Research. 660 Phase I Clinical Trial of the Flavonoid Quercetin
DMSO on the day of use and supplied in glass syringes. It was administered via the side arm of a polypropylene giving set through which RIMSO 50 (50% volume DMSO and 50% water) OH was being slowly injected. A total volume of 50 ml of RIMSO 50 was injected with each quercetin injection. This was essential to prevent precipitation of quercetin in the giving set. The rate of i.v. injection of quercetin at the starting dose of 60 mg/m2 was initially rapid over 30 s, but at doses above 945 mglm2, it was increased to 5 mm because of pain on injection. When the quercetin injection was completed, the drip was left in situ, and 0 200 ml saline were infused over the succeeding 30 mm. The starting dose of 60 mg/m2 was chosen on the basis of Fig. I Structure of quercetin. a previous normal volunteer study (24). Thereafter, doses were increased according to a modified Fibonacci regimen (Table 2), with three patients treated at each dose level if grade 3 or 4 toxicity was not seen. The maximum tolerated dose was con- in this article describe the results of a Phase I trial of quercetin sidered to have been reached when two of three patients or three with pharmacokinetic evaluation. In addition, because quercetin of six patients experienced dose-limiting toxicity. Dose-limiting is a tyrosine kinase inhibitor, we attempted to demonstrate that toxicity was defined as the production of grade 3 or 4 general a pharmacodynamic effect occurred in treated patients. toxicity, or grade 2 renal toxicity, cardiac toxicity, or neurotox- icity. PATIENTS AND METHODS Pharmacokinetics. Venous blood samples for pharma- Patients. Fifty-one patients were entered into this study, cokinetics were drawn at frequent intervals, placed in Li-heparmn the eligibility criteria was: microscopically confirmed diagnosis tubes, and within 15 rain separated by centrifugation. The of cancer no longer amenable to standard therapies; age 18-75 plasma was aspirated, frozen at -70#{176}C, and analyzed within 2 years; performance score of 2 on the WHO scale; life expect- weeks. In one patient, a sample of ascites was drawn before and ancy of at least 12 weeks; informed consent; no previous che- 60 mm after quercetin administration. Urine was collected for motherapy or immunotherapy in the preceding 3 weeks or 6 24 h before and after quercetin administration and stored at weeks for nitrosoureas, mitomycmn C, or extensive radiotherapy; -70#{176}Cbefore analysis. hemoglobin > 10 glliter, WBC count >4 X l09/liter, and plate- HPLC Determination of Quercetin in Plasma. Chro- let count >100 X lO9lliter; and reasonable serum biochemistry matographic analysis of flavonoids was performed using a Kon- with bilirubin <25 p.M. liver enzyme levels less than twice the tron HPLC system (Watford, United Kingdom) with a upper limit of normal, and serum creatinine 120 p.M. All p.-Bondapak C18 column equipped with a high-pressure mixing patients gave written informed consent, as approved by the solvent delivery system (HPLC 422), an automatic sample in- Ethics Committee of the Queen Elizabeth Hospital (Birming- jector (HPLC 465), and diode array spectrophotometric detector ham, United Kingdom). set to 375 mm (HPLC 440). System control, data collection, and Baseline studies before administration of quercetin includ- data evaluation were performed using an IBM PC with a 450 ed: history and physical examination, blood pressure, height, MT2 software data package (Kontron, Watford, United King- weight, performance status, blood count with differential, pro- dom). Standards and control samples were analyzed as previ- thrombin time, serum chemistry including bilirubin, alkaline ously described (25). phosphatase, aspartate aminotransferase, creatinine, urea, so- Pharmacokinetic Modeling. Quercetin plasma levels dium, and potassium, urinalysis for protein and sugar, electro- were modeled to a two-compartment open model using the cardiogram, chest X-ray, and, when appropriate, tumor mea- Statis package based on the Marqhardt algorithm. The data fitted surements. While on study, patients had a weekly review with better to a dual- than to a single-compartment model as judged clinical examination, blood count, and serum chemistry estima- by Akaike and Swartzmann’s criteria. The parameters derived tion. In the first part of this trial, patients were treated at 3-week from the model (A, a, B, and 3) allow calculation of drug intervals. Because myelosuppression was not evident with the clearance (mI/mn), volume of distribution at steady state (li- 3-week schedule, after the dose-limiting toxicity had been en- ters), and AUC. countered, weekly treatments were introduced to intensify drug Determination of Tyrosine Kinase Inhibition in Lym- delivery. phocytes. Twenty ml blood were taken at time intervals up to Quercetin Administration. Quercetin dihydrate was 16 h after quercetin administration. Blood specimens were im- supplied as a pale yellow powder of >99% purity by the Aldrich mediately placed on ice and within 30 mm layered onto 10 ml Chemical Company. It was formulated under sterile conditions ice-cold Lymphoprep (GIBCO) and centrifuged for 30 rain at in a laminar flow cabinet in analytical grade DMSO (Sigma). 500 X g at 4#{176}Cto separate the lymphocytes. The lymphocytes Quercetin was reconstituted at a concentration of 50 mg/ml for were aspirated, added to 40 ml ice-cold wash buffer [50 mt i doses up to 945 mgim2 and at a concentration of 100 mg/mI for Tris-HCI (pH 7.4), 1 misi EDTA, 1 mM EGTA, 150 mr NaCl, higher doses. HPLC studies showed that quercetin remains and 100 p.M sodium o-vanadate], and centrifuged at 500 X g at stable when stored in ethanol or DMSO at 10 nmi over 21 days 4#{176}Cfor 10 mm. The supernatant was removed, and the pellet under air at 4#{176}C(data not shown). Quercetin was made up in was resuspended in another 20 ml ice-cold wash buffer and
Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 1996 American Association for Cancer Research. Clinical Cancer Research 661
Table 1 Patient characteristics Table 2 Number of patients and courses at each dose level Total patients S 1 Quercetin dose No. of No. of 3-Week treatment 37 Dose level (mg/rn2) patients treated courses Weekly treatment 14 1 60 3 9 Male:female 26:25 2 120 3 8 Median, yr (range) 55 (23-78) 3 200 3 5 Histological diagnosis 4 300 3 7 Large bowel 14 S 420 3 8 Ovarian 10 6 630 5 10 Pancreas 7 7 945 8 11 Melanoma 6 8 1400 5 8 Stomach S 9 2000 1 1 Hepatoma 3 10 1700 3 3 Non-small cell lung cancer 2 1 1 (weekly) 1400 8 18 Renal 2 12(weekly) 945 6 8 Other 2 Previous chemotherapy 40 Previous radiotherapy 9 No previous chemotherapy or radiotherapy 10 Performance status renal toxicity was encountered, quercetin was given without 0 6 prehydration. There was no hematological toxicity, neurological 38 toxicity, mucositis, or alopecia. 2 6 3 Toxicity during Bolus Injection. At the first dose level, the only toxicity was mild local pain on injection, which lasted for the duration of injection and for perhaps 10 s afterwards. Following injection there was a brief period of flushing that affected the whole body and was associated with sweating. This recentrifuged as described above. The supernatant was then lasted for approximately 3 mm, but was not associated with any removed, and the pellet was resuspended in 500 p.1 lysis buffer fall in blood pressure. [50 mM Tris-HC1 (pH 7.4), 1 mt i EDTA, 1 ms EGTA, 100 p.M At and above quercetin doses of 945 mgim2, the concen- sodium o-vanadate, 100 p.M phenylmethylsulfonyl fluoride, 25 tration of quercetin injected was increased from 50 to 100 mM benzamine, 50 p.g/ml aprotonin, S p.giml leupeptin, and 2% mg/mi. This was done in an attempt to minimize the volume of Triton X-lOO] and incubated on ice for 15 mm prior to being DMSO injected. It was immediately apparent that at this dose centrifuged at 13,500 rpm for S mm. Lymphocyte lysates were level there was more severe pain on injection, which was local- stored at -80#{176}C. ized to the muscles in the upper arm on the side of the injection. Lymphocyte lysates from an individual patient had protein In an attempt to minimize this, the duration of quercetin slow content estimated to ensure equal loading of lanes and then were bolus injection was increased to S mm, and 10 mg morphine subjected to SDS-PAGE (8% acrylamide gel). Proteins were elixir were given before drug administration. Although no for- transferred onto polyvinylidene difluoride membranes (Milli- mal assessment of the pain score was made, this seemed more pore) over 4 h at 95 mV in blotting buffer comprised of 25 mi i tolerable and decreased the subjective intensity of pain and Tris-HC1 (pH 8.3), 192 iiiM glycine, and 1% SDS in 20% (viv) flushing. methanol. After the transfer, nitrocellulose filters were washed To investigate whether bolus injection of the vehicle in PBS-Tween, and blocked for 2 h at 37#{176}Cinblocking buffer caused flushing and pain, two patients received 50 ml RIMSO containing 5% fat-free dried milk in PBS-Tween (without 50 solution by rapid injection. These patients did not experience azide). The membranes were then incubated with the horserad- any side effects, and it was therefore concluded that quercetin ish peroxidase-conjugated polyclonal antibody RC-20 (Affiniti) and not the vehicle was responsible for pain and flushing on at 4#{176}Covernight. After washing with PBS-Tween, the strips slow bolus injection. Quercetin administered to a patient at the were incubated in detection solution containing 50% luminoll 630-mgim2 dose level via a Hickman catheter resulted in no 50% Enhancer (Amersham ready-to-use ECL system) for 1 mm. pain but slight flushing on injection. The strips were then arranged in a transparency and exposed to During the dose escalation at 1400 mg/m2, patients also Hyperfilm-ECL (Amersham) for 40 mm. reported dyspnea during and for a few minutes after quercetin administration. This was difficult to quantify, but the patient RESULTS treated at 2000 mg/m2 had severe dyspnea which lasted S mm Patient Population. Table 1 lists the characteristics of following quercetin administration. We considered this unac- the patients treated in this study. The median age was 56 (range, ceptable and treated the next cohort of patients at 1700 mg/m2, 23-78) years. The tumor histologies of the treated patients were: at which dyspnea was less prominent. large bowel, 14; ovary, 10; pancreas, 7; melanoma, 6; stomach, Emesis. At the 10th dose level (1700 mg/m2), grade 4 5; renal, 2; hepatoma, 3; and non-small cell lung, 2. Two emesis was encountered. This occurred within minutes of quer- patients had other diagnoses: one with testicular teratoma and cetin administration, which has been the pattern in all patients one with gastrinoma. with quercetin-imduced vomiting. Subsequently, the 5-HT3 an- Dose Escalation and Toxicity. Table 2 shows the dose tagonist granisitron (3 mg) combined with dexamethasone (8 levels and number of patients treated at each dose level. Until mg) were given i.v. before quercetin. Comparing vomiting at
Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 1996 American Association for Cancer Research. 662 Phase I Clinical Trial of the Flavonoid Quercetin
3 Quercetin-induced emesis” Table 4 Renal toxicity and dose level CTC vomiting grade CTC nephrotoxicity (creatinine) Dose level” With iv. granisitron (mg/m2) 0 1 2 3 4 Without iv. antiemetics and dexamethasone Dose level 630 4 0 0 1 0 945 3 3 1 1 - (mg/rn2) 0 1 2 3 4 0 1 2 3 4 0 1400” 3 1 0 0 0 630 8 1 1 0 0 2000 1 0 0 0 0 945 2 0 2 0 0 8 3 0 0 0 1700 0 1 1 0 1 1400 4 1 0 0 0 6 4 2 2 1 l400(weekly)c 3 1 1 0 1 1700 0 0 1 0 1 0 0 1 0 0 945 (weekly) 5 0 1 0 0 2000 1 0 0 0 0 ‘, No renal toxicity was encountered at dose levels below 630 “ The number of observations refers to courses for which data were mg/rn2.
obtained and amounts to 39 of 59, or 66%, of all potentially evaluable 1’ One patient not evaluable. courses. C Two patients not evaluable.
doses of 945 mgim2 and above, without antiemetics, 4 of 12 courses were associated with grade 2-4 vomiting versus 6 of 27 24-h creatinine clearances before and immediately after querce- with antiemetics. Thus, the antiemetics had no significant effect tin therapy. For 1 1 courses at 945 mgim2 and for 2 at 1400 (Table 3). mgim2, the mean pretreatment creatinine clearance was 69.5 Cardiovascular Toxicity. One patient with a history of mlimin, falling to 56.6 mllmin after quercetin treatment (P = hypertension had renal impairment (grade I), but her hyperten- 0.032, paired t test), i.e., a 19 ± 8% fall. Since no myelosup- sion was grade 3 with a blood pressure of 230/130 mm Hg when pression was seen with the 3-week schedule, the dose of 1400 renal impairment ensued. Her pretreatment serum creatinine was mg/rn2 was explored in a more intensive weekly schedule. All of 94 p.M and creatinine clearance was 69 ml/min. In the 24 h after the data given are for treatments given without iv. hydration. treatment, this fell to 46 ml/min, and I week later her serum The first patient treated on the weekly schedule at this dose had
creatinine was 234 p.M and creatinine clearance was 26 ml/min. an unresectable pancreatic cancer. Following quercetin therapy, Within 2 weeks, her creatinine clearance had normalized to 72 he had prolonged nausea and developed acute renal failure mI/mm, and her hypertension had been controlled with atenolol. (grade 4) requiring peritoneal dialysis. Another two of seven Two other cardiovascular events occurred, both in patients patients treated with this dose level on the weekly schedule had treated at 945 mgim2. One patient with a previous history of renal toxicity: one patient had grade 1 and one patient had angina developed central chest pain 24 h after the second course grade 2. of quercetin and died on the second day after treatment. Another Four patients on the weekly schedule had three or more patient developed grade 3 nephrotoxicity after the second course courses of quercetin. Fig. 2B shows pretreatment creatinine of quercetin, developed back pain, and died in heart failure 16 clearance versus day on study. There was no cumulative fall in days after the second course. A postmortem examination re- creatinine clearance with multiple weekly treatments, indicating vealed aortic dissection extending beyond the renal arteries. that any nephrotoxicity due to quercetin was transient and Renal Toxicity of Quercetin. Three patients were reversible within 7 days. treated at 1700 mg/rn2, all of whom experienced renal toxicity, Overall, 2 of 10 patients who were treated at 1400 mgim2 with elevation of serum creatinine (Table 4). These patients also had clinically significant nephrotoxicity (1 patient with grade 4, experienced nausea and vomiting within minutes of the querce- and 1 patient with grade 2). Since this had occurred with the first tin injection. In one patient (Q26) treated at 1 700 mgim2, nausea course of quercetin, we therefore enrolled an additional six was grade 4 (Table 3) and more prolonged. This may be because patients at the preceding dose level of 945 mg/rn2 weekly. At this patient developed grade 4 renal toxicity, with serum creat- 945 mgim2, eight patients were treated at 3-week and 6-week mine reaching a peak of 2145 p.M I I days after treatment, which intervals. Overall, combining data for both schedules, 3 of 14 returned to normal by day 34 (Fig. 2A). Two other patients were patients developed clinically significant renal impairment, 2 treated at this level: one experienced grade 2 renal toxicity and developed grade 2, and 1 developed grade 3. the other grade 1 toxicity, but in patient Q27 the serum creati- Abrogation of Renal Toxicity with i.v. Hydration. One nine remained elevated at 246 p.M on day 34 posttreatment, and patient treated at 630 mg/m2 developed grade 3 renal toxicity, in patient Q28 the serum creatinine was elevated at 149 p.M on with elevation of the serum creatinine to 420 p.M by day 10. By day 28. None of the patients who experienced renal toxicity had day 21 this patient’s serum creatinine had normalized (Fig. 2C). evidence of nephritis, infection, or obstructive uropathy on This patient was keen for additional therapy and was therefore ultrasound scanning of the abdomen. retreated, but with prehydration of 1 liter normal saline given Dose Escalation with Weekly Treatment and Nephro- over I h, and after quercetin treatment, 0.5 liter 5% dextrose was toxicity. At the highest doses of quercetin administered (2000 given; there was no fall in creatinine clearance. The next course and 1700 mgim2), myelosuppression was not seen, but there in this patient was given at 945 mg/rn2 with hydration, and no were dose-limiting toxicities. We therefore attempted to dose renal toxicity ensued. In other patients, hydration was given intensify by using a weekly schedule. During this weekly sched- before bolus injection of quercetin at 945 mg/rn2, and the ule, a concerted effort was made to monitor renal function with pretreatment clearance was 82.3 ± 6.7 mI/mm. In the 24 h after
Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 1996 American Association for Cancer Research. Clinical Cancer Research 663
i 10000 1 -#{149}-- Ursa -- 0- - Creatlnlne 1700 #{163}: -Q 1000 0 0.