Intraperitoneal Drug Delivery for Ovarian Cancer: Why, How, Who, What, and When?
Intraperitoneal Drug Delivery for Ovarian Cancer: Why, How, Who, What, and When?
ABSTRACT: In 1996, intraperitoneal (IP) administration of cisplatin plus intravenous (IV) cyclophosphamide proved superior to both drugs given intravenously at the same doses—which, at the time, was the standard treatment in the United States. The IP ‘option’ was not adopted, however, because the standard treatment had shifted to IV cisplatin plus paclitaxel. Two additional phase III trials by the Gynecologic Oncology Group (GOG) comparing IP versus IV cisplatin, but including other variables, have shown similar superior effects of the IP route on outcome, but with toxicities—particularly local tolerance and neuropathy—increased. An ongoing trial by the GOG is again looking into an IP versus IV comparison, and introducing in one of the IP arms the substitution of IP carboplatin for IP cisplatin. All three arms of this trial contain bevacizumab (Avastin). Two other trials comparing IV versus IP administration of platinums or platinums and paclitaxel have just been launched, led by Japanese and Canadian investigators, respectively. While awaiting additional data on the ongoing debate over IP versus IV therapy, it is important that we consider issues concerning why the IP route may be relevant, how can one increase the safety of this route, and who should be treated and with what drugs, particularly when faced with a patient outside the clinical trials setting. The underlying hypothesis for use of IP therapy is based on the existence of a dose-effect relationship for platinum drugs in ovarian cancer. We review the known data on this relationship, and explore why interest in platinum drugs has become the central focus of ovarian cancer treatment.
Epithelial ovarian cancer (EOC) spreads prominently within the peritoneal cavity. In fact, we now know that high-grade serous cancers are often of tubal origin, and their presentation as tubo-ovarian masses renders it likely that intraperitoneal spread occurs as an early event in their clinical evolution. From the outset, IP drug administration in ovarian cancer conferred a pharmacologic advantage over IV administration, given the additional clinical benefit derived from achieving more efficient control of life-threatening peritoneal disease.
The systemic treatment of epithelial ovarian cancer and related extrauterine adenocarcinomas of Mullerian origin relies on ‘platinum-based’ chemotherapy. Ever since the introduction of cisplatin into our armamentarium in the late 1970s, this drug, and later its less toxic analog, carboplatin, have provided the platform for treatment of these malignancies at all stages. Adoption of cisplatin as standard treatment accelerated with the discovery of effective antiemetics, and introduction of the equally efficacious carboplatin, with its more predictable pharmacokinetics and superior safety profile, led to wide exploration of dose-response relationships.[3,4] Before meta-analyses were in vogue, William Hryniuk and colleagues observed that in ovarian cancer trials, the dose of cisplatin correlated with improved survival. However, evidence for benefit beyond a certain ‘dose-intensity’ was not obvious. As Martin Gore noted in a 2003 editorial in the Journal of Clinical Oncology, “Devotees of the ‘more is better’ school of oncology must face the uncomfortable truth that the majority of randomized trials have failed to show an overall survival benefit associated with an increase in the total dose or dose-intensity of platinum.” He went on to state, however, “Any discussion of dose-response in ovarian cancer is incomplete without a reference to the subject that always guarantees a thoroughly entertaining argument: namely, intraperitoneal chemotherapy. The data from randomized trials of intraperitoneal chemotherapy do support the premise that a dose-response exists....”
This discussion is still applicable from today’s perspective, but it has become even more relevant than when Gore’s editorial was written. In 2006 the US National Cancer Institute (NCI) issued a clinical announcement after the results of the third of phase III GOG trials (GOG172) on IP versus IV became known and were subsequently published (by Armstrong, in the New England Journal of Medicine, in 1996, and others).[9-11] After reviewing the data from the IP versus IV phase III clinical trials conducted by the GOG (Table 1), the NCI announced, “Based on the results of these randomized phase III trials, a combination of IV and IP administration of chemotherapy conveys a significant survival benefit among women with optimally debulked epithelial ovarian cancer, compared to IV administration alone.” Further, the NCI urged physicians to become familiar with these trials and, if necessary, refer patients to centers familiar with IP drug delivery. Both in the US and Europe, reaction to such announcement was swift: Ozols and colleagues noted that “…exploratory cross-trial comparisons of IV carboplatin/paclitaxel compared to IP regimens suggest very similar efficacy in optimal stage III disease,” and Gore et al argued, ‘There are at least eight reasons why the Armstrong article does not support the use of IP administration as a standard of care.”
A major problem in the adoption of IP therapy as the new standard for optimally debulked ovarian cancer might be put forth in a format analogous to the title of Gore’s 2003 editorial in JCO: “IP Cisplatin Is More Effective than IV Cisplatin: But How Do I Prescribe It?” Figure 1 shows the schema of GOG172 and Figure 2 graphically lays out many of the disputed areas in GOG172, including: Is IP paclitaxel necessary to achieve the results? Is the dose of IP cisplatin of 100 mg/m2 acceptable to many patients? How does one deal with the 24-hour infusions of paclitaxel? If one had used a carboplatin control, would the survival differences have been significant? Can one advocate a treatment with a 42% completion rate? And, is the tradeoff of adverse quality of life for a few months and risk of treatment complications from IP therapy for a less than 20% reduction in risk of dying reasonable? GOG investigators struggled with these issues, and after a series of pilot studies embarked in late 2009 on another IP versus IV phase III study (GOG252) containing also an IP toxicity–reduction arm, based on IP carboplatin and phase II data from Japan.
With this as a background, we expand on this debate and provide some additional perspective: the effect of platinum drugs on the outcome of ovarian cancer is even more relevant today than it was in the past, given the apparent shortcomings of many new drugs that have been tested. The route of platinum administration, as well as patient selection and the role of other drugs, should be central to ongoing research in this area.
How? An Outline of Methods for IP Drug Delivery
IP drug delivery is an old topic. The fact that the administration of IP chemotherapy is more complex and time-consuming than IV chemotherapy, has been associated with a greater risk of toxicity, and is both physically and emotionally more challenging for patients continues to shape the nonacceptance of this modality by clinicians. There is also a learning-curve in acquiring the proper techniques for surgically placing an IP port and developing the clinical expertise in accessing the IP port for drug delivery. There continue to be debates beyond the scope of our expertise regarding whether catheters with multiple holes or a single one should be used and the location of their placement. Also, insistence on heating solutions and making patients roll during the infusion seems misplaced. GOG institutions continue to accrue in a satisfactory way to the current trial, as they did in the past. GOG172 did have a greater problem in completion rate relative to the two prior GOG phase III trials, and also a greater number of catheter problems.[11,15,16] This higher complication rate in the last trial has been attributed to the greater aggressiveness of the debulking surgery performed, but is likely also due to the added IP access (twice every 3 weeks) and local toxicity that occurs with addition of IP paclitaxel.
From the extensive experience in past studies, some observations on use of the IP route included the following:
• Multiple surgeries increase the incidence of catheter malfunction;
• Certain vesicant drugs (eg, mitoxantrone) increase catheter malfunction[17,18];
• Drug delivery is generally more satisfactory in patients selected for IP therapy based on low-volume disease (Figure 3);
• Repeated percutaneous drug administration, while feasible, does not have a satisfactory record of completion (example: GONO trial);
• At least for recurrent disease, volumes greater than 1 cm do not respond to any extent to IP drug administration (and the rationale for greater delivery of IP drug over IV administered drug to those tumor nodules is weak); and
• Single institutions with sufficient experience in IP drug delivery have a low incidence of complications from IP drug administration.
While these ‘dogmas’ should be appropriately challenged by new data, they should still serve as general guidelines for satisfactory IP drug delivery.
From years of experience, certain procedural details emerge as important. Nurses experienced in the administration of IP chemotherapy are essential; they have the technical expertise in managing the “mechanics” of IP drug delivery and can meet the challenge of helping patients cope with the physical and emotional demands of undergoing IP therapy. Any problem with gravity administration of IP chemotherapy is a signal to reinspect whether the IP catheter is patent. If it is uncertain, rather than wasting precious time in clinical facilities, it is best to have a radiologist determine whether the catheter placement is correct and uncover any reasons for the inflow obstruction (usually loculation or complete occlusion, causing backflow around the cather and into the subcutaneous tissues). Regional side effects such as pain and bloating from abdominal distension are common problems with the initial delivery of chemotherapy via the peritoneal route. All women do not tolerate the same amount of IP fluid volume: a rough guideline we have adopted is using 1 L/m2.
Outflow obstruction is common: it is probably related to a small amount of connective tissue within the rim of the outflow orifice. In phase II/pharmacologic studies, not infrequently we collect specimens (including cytology) if outflow is satisfactory. The longer an IP catheter is left in place, however, the percentage of outflow obstruction rises to above 40%. If it is desirable, it is best to maintain the patency of the IP catheter by flushing the system with heparinized saline every 6 weeks. Nowadays, IV bevacizumab is often coadministered and continued beyond IP therapy, raising the new question of when is it most desirable to remove the port and catheter under such circumstances. We did observe an episode of chemical peritonitis, when IV pegylated liposomal doxorubicin was continued in a patient who had not undergone removal of a previously placed IP catheter (FM, personal observation).
Who Has Been Shown to Benefit From IP Therapy?
Single-institution reports of long-term survivors following failures of IV drug administration provided initial support for IP drug administration[24-31] and eventually spurred proof-of-concept randomized trials. These initial IP phase I/II and pharmacologic studies were reported throughout the 1980s, suggesting that survival was prolonged in patients who received salvage IP therapy after initial chemotherapy and second-look surgery (with minimal residual disease). Eventually, randomized studies provided the key evidence for efficacy applicable to patients diagnosed with high-grade serous epithelial ovarian cancer (and those of tubal and primary peritoneal origin) that predominantly spread to the omentum and the peritoneal cavity by intraabdominal dissemination (thus, stage III), and also to pelvic and paraaortic nodes by lymphatic drainage (locoregional disease).
Eight randomized controlled trials have assessed IP therapy versus IV chemotherapy for first-line treatment in EOC. While the three studies performed by GOG (Table 1) encountered substantial criticisms, they were adequately powered and the differences shown are persuasively linked to the assigned route of administration. In the five trials shown in Table 2, there were no statistically significant differences, but the studies were uninterpretable because of lack of statistical power, in some clearly attributed to their early termination. The inability to adequately conduct these studies is a reflection of some of the inherent difficulties faced by clinical investigators involved in testing IP therapy.
After the NCI announcement, many groups in the US adopted GOG172 or its modifications for stage III optimally cytoreduced ovarian cancer. In the subsequent years, use of such IP therapy outside a trial setting may actually have declined, mainly because of difficult patient acceptance and competing clinical study options. For US GOG institutions, decisions have now been simplified because of availability of GOG252 as a treatment choice, until 2013. Who should be receiving IP therapy beyond the confines of a clinical trial? And paraphrasing Martin Gore, how do we prescribe it? Some historical perspective is helpful: when IP therapy became widely applied in the 1980s within the US, a number of phase II studies disclosed results focusing primarily on response rates determined by surgical reassessment—an inappropriate endpoint for patients who have primarily low-volume disease.
The GOG, in fact, opened a series of drug studies (GOG102 series) that were mostly inconclusive, except for finding that platinum sensitivity was a determinant of outcome. Another determinant of outcome was the volume of disease (bulk disease may, in fact, correlate to some extent with the absence of platinum sensitivity). After years of uncontrolled series within institutions and cooperative groups, Robert Ozols in 1991 wrote an editorial entitled “IP Therapy of Ovarian Cancer: Time’s Up,” and certainly it was about time to move on from these types of trials. Fortunately, since June 1986, the Southwest Oncology Group (SWOG) under David Alberts’ leadership had started an IP versus IV trial in the population most likely to be sensitive to cisplatin: previously untreated patients. It later became an intergroup study (GOG104), with GOG enrolling 295 patients, 298 enrolled by SWOG, and 61 by the Eastern Cooperative Oncology Group. In January 1991, with no knowledge of outcome, an amendment was passed to expand the study by 1 year “to achieve a large enough sample for a separate analysis of data from patients with residual tumors that were no larger than 0.5 cm.” Such an amendment was later a source of much criticism (see below). The study completed accrual in July 1992 and was published in December 1996.
This first phase III trial showed an advantage for progression-free survival (PFS) and overall survival (OS) for the IP arm. Criticisms would follow this trial and would further escalate upon publication of the two subsequent phase III trials, also summarized in Table 1. While many of the criticisms are valid, including the multiple variables introduced into the IP versus IV question in subsequent trials, the common thread in the superior results is the IP administration of cisplatin. The three GOG trials taken together lead to the following perspective: 1) the IP route for cisplatin was favored in all three trials; 2) the results were achieved in previously untreated patients, when they would be expected to be most sensitive to platinums; and 3) IV carboplatin, while outperforming cisplatin (nonsignificantly) in some trials, is unlikely to yield significantly superior results over IV cisplatin (and why would one expect ovarian cancer to be different from any other neoplasm in this regard?). With respect to the greater toxicity of IP therapy, two factors are obvious contributors: 1) the dose of IP cisplatin being higher than the IV dose in the last two trials is a major contributor to the greater intolerance and would be expected to improve by a decrease to 75 mg/m2 (well documented in IV studies); and 2) IP paclitaxel, by accessing the peritoneal cavity twice as often, carries higher toxicity, and the drug itself adds to both local and systemic toxicities (especially neuropathy). Diminishing the dose of cisplatin has widely been adopted; however, eliminating IP paclitaxel meets with resistance by ‘pure’ trialists. Given that two studies omitting IP paclitaxel led to efficacy similar to that seen in the third trial, in our view superiority of IP paclitaxel regimens to either cisplatin or carboplatin alone given by the IP route must be demonstrated before there can be wider adoption of the GOG172 modified regimen.
IP treatment, nevertheless, should be considered for patients of similar stage and residuum who had been included in the three GOG trials. GOG104 included patients with < 2 cm residual disease—a somewhat more liberal entry criterion, explaining in part the lesser median survival time observed relative to the later trials. However, significant benefit of IP over IV was not demonstrated in the subset with smallest volumes of disease, indicating that the patients with larger residuum contributed to the outcome. This could argue in favor of not needing to be so strict with optimal cytoreduction—a topic we take up later in this article. It is noteworthy that GOG114 has the best PFS result of all the three trials, and provocatively this could lead to the interpretation that ‘chemical cytoreduction’ by IV carboplatin has rendered the efficacy of IP cisplatin more demonstrable. At present, one would continue to consider the best candidates for IP therapy those who are previously untreated and optimally cytoreduced.
Finally, what if one introduces a biomarker of platinum sensitivity to the selection of patients for IP cisplatin? Krivak and coworkers have used BRCA1 protein immunostaining as a surrogate for platinum sensitivity (with low BRCA1 expression occurring in about 50% of patients entered into GOG172, being considered as indicative of greater sensitivity to platinum). In their retrospective study, 93 patients with low BRCA1 expression who received IV treatment were compared with 96 patients, also with low BRCA1 expression, who received IP treatment. Their median PFS times were 18.7 and 34.7 months, respectively, and their median OS times were 47.7 and 84.1 months, respectively. There were no differences in PFS or OS for patients treated by the IV versus IP method in the BRCA1 high-immunostaining patients. Certainly, if validated, this biomarker could greatly enhance the applicability of IP therapy by leading to the proper selection of patients. Also, the results are noteworthy on theoretical grounds: if the most sensitive patients benefit, this tells us that it is the platinum dose-intensity achieved by the IP route and not the relatively small increments achievable via the IV route that determine outcome. Conversely, even a greater than 10-fold increase in platinum exposure may be futile if platinum resistance is present.