Objective
Treatment for advanced-stage epithelial ovarian cancer (AEOC) includes primary debulking surgery (PDS) or neoadjuvant chemotherapy (NACT). A randomized controlled trial comparing these treatments resulted in comparable overall survival (OS). Studies report more complications and lower chemotherapy completion rates in patients 65 years old or older receiving PDS. We sought to evaluate the cost implications of NACT relative to PDS in AEOC patients 65 years old or older.
Study Design
A 5 year Markov model was created. Arm 1 modeled PDS followed by 6 cycles of carboplatin and paclitaxel (CT). Arm 2 modeled 3 cycles of CT, followed by interval debulking surgery and then 3 additional cycles of CT. Parameters included OS, surgical complications, probability of treatment initiation, treatment cost, and quality of life (QOL). OS was assumed to be equal based on the findings of the international randomized control trial. Differences in surgical complexity were accounted for in base surgical cost plus add-on procedure costs weighted by occurrence rates. Hospital cost was a weighted average of diagnosis-related group costs weighted by composite estimates of complication rates. Sensitivity analyses were performed.
Results
Assuming equal survival, NACT produces a cost savings of $5616. If PDS improved median OS by 1.5 months or longer, PDS would be cost effective (CE) at a $100,000/quality-adjusted life-year threshold. If PDS improved OS by 3.2 months or longer, it would be CE at a $50,000 threshold. The model was robust to variation in costs and complication rates. Moderate decreases in the QOL with NACT would result in PDS being CE.
Conclusion
A model based on the RCT comparing NACT and PDS showed NACT is a cost-saving treatment compared with PDS for AEOC in patients 65 years old or older. Small increases in OS with PDS or moderate declines in QOL with NACT would result in PDS being CE at the $100,000/quality-adjusted life-year threshold. Our results support further evaluation of the effects of PDS on OS, QOL and complications in AEOC patients 65 years old or older.
See related editorial, page 695
Epithelial ovarian cancer continues to be the most lethal gynecological cancer, with an estimated 21,980 new cases and 14,270 deaths in the United States in 2014. The high mortality rate is largely due to the majority of patients being diagnosed at International Federation of Gynecology and Obstetrics stage III or IV. Primary debulking surgery (PDS) followed by chemotherapy is the standard of care for patients diagnosed with advanced epithelial ovarian cancer (AEOC).
Studies have demonstrated the importance of optimal debulking (<1 cm residual disease) combined with the timely administration of chemotherapy as important predictors for improved overall survival (OS). Survival is significantly worse in patients in which surgery results in suboptimal debulking, and evidence does not demonstrate a survival benefit for a secondary debulking procedure when an adequate attempt at primary debulking was undertaken at the first surgery. Reported optimal cytoreduction rates vary between 15% and 85% because of many factors including surgeon experience and reporting, volume of disease, location of disease, and/or patient performance status.
Neoadjuvant chemotherapy (NACT) prior to debulking surgery has been proposed for some patients to increase the feasibility and success of optimal surgical cytoreduction and decrease morbidity without compromising survival. Vergote et al performed a randomized controlled trial (RCT) (European Organisation for Research and Treatment of Cancer [EORTC] 55971) in several European countries comparing PDS with NACT for patients with AEOC. NACT was found to be noninferior when compared with PDS for OS and progression-free survival with lower adverse events and morbidity.
Given our aging population, of particular interest is the treatment of elderly patients with AEOC. Several studies have shown complications from aggressive surgery are higher in this population. In 2011, Wright et al published a retrospective study evaluating the complication rates of radical debulking surgery by age and surgical complexity. Age and number of procedures were shown to be the strongest predictors of morbidity and mortality with a linear increase in risk seen across age categories among all types of complications.
Moore et al retrospectively evaluated patients over the age of 80 years with AEOC. This study showed that patients in this age group had a disproportionately high rate of complications and postoperative deaths with cytoreductive surgery, with 20% of patients dying from postoperative complications within sixty days of surgery. Furthermore, many elderly patients receive either limited or delayed chemotherapy and may even be unable to receive chemotherapy after PDS, with patients over the age of 80 years receiving a complete course of initial chemotherapy less than 20% of the time.
We sought to create a theoretical model to evaluate the cost-utility implications of treatment with NACT vs treatment with PDS among patients over the age of 65 years diagnosed with AEOC.
Materials and Methods
Model
We created a Markov transition state model using Tree Age software (TreeAge Inc, Williamstown, MA) to compare the cost-utility of 2 initial treatment scenarios for a hypothetical cohort of women over the age of 65 years newly diagnosed with AEOC (stages III and IV). Arm 1 was modeled as patients treated with PDS followed by 6 cycles of chemotherapy with carboplatin and paclitaxel (CT) as described in Gynecologic Oncology Group (GOG) 158. Arm 2 was modeled as patients treated with 3 cycles of NACT with CT followed by interval debulking surgery and then an additional 3 cycles of CT with dosing equivalent to patients in arm 1 ( Figure 1 ). Markov states were treatment phase, recovery phase, survival, and death. The cost and quality of life (QOL) impact of acute surgical complications were modeled during the postsurgical recovery phase.
The model parameters included OS, surgical complications, probability of initiating secondary treatment (eg, chemotherapy following primary debulking), treatment cost (including cost of chemotherapy, surgery and surgical add-on procedures, and hospitalization to account for surgical complications), and QOL estimates during each model state.
Based on the EORTC 55971 study, OS (median of 29 months) and QOL during the progression-free survival stage were assumed to be equal. Surprisingly, despite significant differences in surgical complications across the 2 groups, patients reported similar quality of life in both groups. Parameter estimates and sources are listed in Tables 1-3 . A 5 year Markov state transition model with 3 month cycles was created to evaluate the cost-effectiveness of NACT relative to PDS. Cost implications were analyzed from the perspective of the health care system.
| Probabilities | PDS | NACT | Source |
|---|---|---|---|
| Probability (range) | Probability (range) | ||
| Receiving chemotherapy | 0.943 (0.90–1.0) | 1.0 | EORTC RCT |
| Receiving surgery | 1.0 | 0.905 (0.85–0.95) | EORTC RCT |
| Major complications | 0.513 (0.26–0.75) | 0.394 (0.20–0.59) | Thrall et al |
| Minor complications | 0.250 (0.13–0.38) | 0.252 (0.13–0.38) | Thrall et al |
| Discount rate | 0.03 (0.00–0.05) | 0.03 (0.00–0.05) | |
| Overall survival | 29 months (15–44) | 29 months (15–44) | EORTC RCT |
| Quality of life | Utility index (range) | Utility index (range) | |
| Active treatment | 0.791 (0.40–0.84) | 0.791 (0.40–0.84) | GOG 152 |
| Immediate recovery | 0.779 (0.38–0.84) | 0.779 (0.38–0.84) | GOG 152 |
| Ongoing recovery (>6 mo) | 0.840 (0.42–0.84) | 0.840 (0.42–0.84) | GOG 152 |
| Costs | US$ (range) | US$ (range) | |
| Chemotherapy | $4939 (2470–7409) | $4939 (2470–7409) | Medicare a |
| Surgery | $2712 (1356–4068) | $1868 (934–2802) | Medicare b |
| Hospitalization/no complications | $10,185 (5093–15,278) | $10,185 (5093–15,278) | HCUP c |
| Hospitalization/minor complications | $16,317 (8159–24,476) | $16,317 (8159–24,476) | HCUP c |
| Hospitalization/major complications | $34,313 ($17,157–51,470) | $34,313 ($17,157–51,470) | HCUP c |
a Medicare reimbursement estimates published by Havrilesky et al
b Medicare reimbursement database maintained by the American Medical Association
a Estimates based on the study by Thrall, et al
b Medicare reimbursement database maintained by the American Medical Association
c Rate of ileostomy and colostomy creation was assumed to occur in the same proportion as small and large bowel resections, given only data available included all ostomies in 1 category
d Splenectomy was used as a surrogate charge for upper abdominal procedures.
| Complication | PDS, % of patients | NACT, % of patients | Classification, minor or major |
|---|---|---|---|
| Transfusion | 6.86 | 5.47 | Minor |
| Surgical injury | 6.55 | 6.42 | 50% minor/50% major |
| Cardiac | 11.52 | 9.19 | Major |
| VTE | 4.75 | 4.85 | 50% minor/50% major |
| General infections | 17.70 | 11.44 | 50% minor/50% major |
| Neurological or renal | 4.25 | 2.25 | 50% minor/50% major |
| Reoperation | 9.93 | 8.49 | Major |
| Pulmonary | 11.23 | 3.81 | Major |
| Wound infection | 20.66 | 14.21 | 50% minor/50% major |
Costs
Overall treatment costs included 3 elements: surgery, chemotherapy, and hospital stay ( Table 1 ). Surgery costs were constructed from national figures using the database of Medicare reimbursement rates maintained by the American Medical Association. The base cost for the NACT interval debulking ($1410) was based on the Current Procedural Terminology code for debulking for ovarian cancer, whereas the PDS base cost ($1971) was based on the Current Procedural Terminology code for radical debulking ovarian cancer surgery to accommodate for the increased surgical complexity of PDS.
The final cost of the surgery was calculated as the base surgical cost plus additional costs for add-on procedures such as bowel resection, ostomy creation, and upper abdominal procedures occurring at the rates previously reported among patients older than 65 years undergoing either primary or interval debulking ( Table 2 ). Specifically, the additional cost associated with the add-on procedures was calculated as a weighted average, taking into account the probability of each additional procedure and the cost associated with that procedure ( Table 2 ).
All PDS patients were expected to receive surgery, and all NACT patients were expected to receive chemotherapy ( Table 1 ). However, based on trial data, we expected some noncontinuation in phase II of treatment. For PDS patients, 9.5% were modeled as not receiving chemotherapy in phase II. For NACT patients, 5.7% were expected not to receive surgery. Chemotherapy costs were based on previously published Medicare estimates for 6 cycles of CT and assumed to be equal in both groups for 6 completed cycles. The variation in chemotherapy costs as shown in Table 4 reflects the noncontinuation rate.
