Do all patients that undergo a ‘complete’ secondary cytoreductive surgery for platinum-sensitive recurrent ovarian cancer, benefit from it?

Outline

Selection criteria for secondary CRS and disease sites and surgical procedures

Apart from the general selection criteria for any surgery and platinum sensitivity, criteria for predicting a CGR were used in two of the three trials.

Perioperative outcomes

Systemic therapies

Survival outcomes

Pre-planned and post-hoc analyses of prognostic factors

Predictive scores for a complete gross resection

The Memorial Sloan-Kettering (MSK) score was the first to be developed (Table 3) [15]. The validated and most widely used score is the AGO score (Table 3), and using the AGO criteria, a CGR was achieved at 76 % in the validation study, but the negative predictive value was 38 % and specificity only 53 % [14]. In one validation study, 63 % with a negative AGO score had a CGR [11]. In a study that used the Tian model (i-model) (Table 3), patients with a low risk of incomplete cytoreduction had a score of 0–4.7, a CGR rate of 53.4 %, and patients with a CGR had a median OS of 63 months, whereas, those with a high risk had a score of >4.7, a CGR rate of 20.1 % and a median OS of 40 months in patients with a CGR [16]. External validation of the Tian model showed sensitivity and specificity values of 83.3 and 57.6 %, respectively [17]. Whereas this score stratifies patients into different prognostic sub-groups, it is unclear if the low-risk group benefits more from surgery than the high-risk group or vice versa. In the SOC-1 trial, patients with an i-model score of 4.7 seemed to benefit the most from surgery (Table 2) [6]. One retrospective study showed a good concordance between the MSK and Tian scores (both accurately predicted a CGR in more than 80 %), while the AGO score was accurate only in 49 % [18]. The AGO score and Tian model have been validated, and the MSK score has not been validated in randomized trials [5, 6]. It has been proposed that the AGO score should be applied first, followed by the MSK score to select patients for surgery [18]. The Tian score or i-model criteria may be used for further stratification in patients with an intermediate MSK score. However, as with the Tian score, none of these scores can point out the magnitude of the benefit from sCRS.

Other criteria for predicting a CGR

For any CRS procedure, surgical decision-making is based on the extent of the disease and the involvement of critical anatomical structures that would preclude a complete cytoreduction or have a detrimental impact on the long-term quality of life. For example, involvement of the porta hepatis in a patient who has had prior surgery in that region could preclude a CGR even if the disease is limited.

Involvement of the bladder trigone would necessitate a cystectomy with or without exenteration, which may not be advocated, especially if there is a peritoneal disease [19]. Another factor that needs to be considered in cases with disease in the small bowel is the length of the remaining small bowel and its impact on the quality of life. The evaluation of disease extent is based on cross-sectional imaging studies with or without the addition of a staging laparoscopy. All three scores predict resectability as a function of other prognostic factors and not based on objective radiological features for resectability. The only radiological findings included are the presence of ascites and multiple disease sites [15, 16]. It may be assumed that imaging findings were considered in decision-making. In a multi-institutional study on iterative CRS for colorectal peritoneal metastases, a CC-0/1 resection was achieved in 83.5 % of the patients [20].

Future efforts need to focus on reducing the proportion of patients having an incomplete resection, and scores and selection criteria should incorporate imaging features and/or sites of disease.

Should patients not be offered sCRS outside a clinical trial in light of current evidence?

Given that one out of four patients will have an incomplete resection and the possibility of a poorer survival compared to SC alone, should sCRS be offered to patients outside clinical trials?

The NCCN (National Comprehensive Cancer Network) guidelines recommend sCRS as a therapeutic option for patients with platinum-sensitive recurrence if the AGO criteria are fulfilled and a CGR can be achieved [7]. The ESMO (European Society of Medical Oncology)-ESGO (European Society of Gynaecologic Oncology) guidelines recommend sCRS for oligometastatic platinum-sensitive first recurrence [8]. The rate of CGR in clinical trials should not be the benchmark, but the insights gained from these trials should be used for better patient selection, limiting sCRS to only those with a high likelihood of CGR. Other factors that should be considered are discussed below.

Sites of disease and disease extent

None of the three trials uses either the sites of disease or the disease extent as a stratification factor, and information on the impact of this critical prognostic factor is missing. All three included patients with both intra-abdominal and extra-abdominal disease were included. According to the number of lesions, the hazard ratios for PFS in the SOC-1 trial were 0.59 for 1–3 lesions, 0.52 for 4–19 lesions, and 0.57 for >19 lesions. It has not been specified how the number of lesions was considered-three lesions confined to the pelvis cannot be compared to three lesions in different parts of the abdominal cavity (e.g., in the pelvis, mid and upper abdomen) [6]. This information on the disease sites is crucial from the morbidity point of view: one 3 cm nodule on the lesser curve of the stomach cannot be compared to a solitary 3 cm nodule in the pelvis or the abdominal wall. One extrapolation from the trials could be that when the margin of benefit is small, and the anticipated morbidity is high, systemic therapy alone could be a reasonable option. Peritoneal metastases, intra-abdominal lymph nodes, pleural metastases, parenchymal liver metastases, and mediastinal nodes are the most common sites of recurrence/progression that are likely to have different prognosis but are all considered together. In the GOG-213, the worst hazard ratios for surgery were for intra-abdominal disease as compared to liver metastases and extra-abdominal sites of disease [4]. sCRS is a non-curative surgery for most patients, potentially prolonging survival and preserving the quality of life. In patients with limited but chemo-sensitive disease and a long platinum-free interval, can SC induce a complete response and produce the same survival as sCRS followed by SC? If the Gompertzian kinetics is considered, chemotherapy should be more effective on smaller or limited tumor burden [21]. In the GOG-213, the median OS with systemic treatment alone was more than 64 months [4]. Thus, some patients experience prolonged survival with SC and targeted therapies alone. These are crucial questions for decision-making by multidisciplinary teams and for conveying accurate prognosis and treatment options to patients.

Isolated recurrence

In one of the first studies in the modern era on sCRS, patients with a single site or isolated recurrence had significantly better OS after sCRS (60 m vs. 42 months for multi-site disease vs. 28 months for carcinomatosis; p<0.001) [15]. This study is nearly 20 years old and included patients treated before 2001, during which the techniques of CRS were being developed. In-field recurrence after peritonectomy could differ from recurrence when no peritonectomy has been performed for the reasons discussed here. First, the peritoneum is the first line of defense, which prevents retroperitoneal tumor extension when not breached [22]. Prior surgery breaks this barrier, and following a peritonectomy, recurrent tumors will infiltrate a varying extent of the retroperitoneum or the underlying parietal and visceral structures. Due to this, even if a CC-0 resection or CGR is performed, it will be an R1 resection as there is a high probability of microscopic residual disease at the resection site. Second, diagnosing recurrence on imaging may be more challenging than exploring the abdomen thoroughly after a previous extensive CRS. Staging laparoscopy may not be able to evaluate all regions [23]. Thus, there could be more disease than is visible, and the recurrence may not be truly isolated. A meticulous evaluation of the disease extent should be performed. MRI peritoneum may be a valuable tool in this regard [24].

Is there a role of sCRS in patients receiving bevacizumab?

In the GOG trial, bevacizumab, in addition to SC, produced a similar or superior survival when compared to sCRS with SC with/without bevacizumab. However, the role of bevacizumab was not considered in patients undergoing a CGR. The results of GOG-213 cannot be interpreted as having no role in sCRS, though the median OS achieved with systemic therapies alone is remarkable [25]. Current treatment guidelines recommend the use of bevacizumab during first-line therapy, which has led to the widespread use of bevacizumab in patients with advanced EOC. Approximately 50 % of patients in high-income countries receive first-line maintenance bevacizumab [26, 27]. In the GOG-213, bevacizumab was used only in 11 % during first-line treatment. Hence, it cannot be assumed that if bevacizumab is reintroduced, it will provide the same survival benefit as when used for the first time [28]. Bevacizumab was used in only 23 % of the patients in the DESKTOP-III trial, and therefore, the benefit of sCRS in patients receiving the drug could not be elucidated [5]. In GOG-213, non-serous and non-clear subtypes benefited from secondary CRS [HR-0.59 (95 % CI 0.21–1.69)] despite using bevacizumab [4]. More evidence is needed to identify which patients could be treated with SC alone when bevacizumab is used and which patients should have sCRS in addition to bevacizumab and SC. The side effects of systemic treatment, the cost of treatment, and the long-term impact on quality of life should be considered.

Is there a role of sCRS in the era of PARP inhibitors?

PARP inhibitors are now recommended as second-line therapy irrespective of the BRCA mutation status [9]. In all three trials, they were used in <20 % of the patients, and therefore, no conclusions can be derived from these trials on the benefit of sCRS in patients receiving PARPi. In a retrospective study of 126 patients treated between 2005 and 2016, there was no benefit in post-recurrence survival with secondary CRS in patients with BRCA1/BRCA2 mutated tumors, while a benefit was in BRCA1/2 wild-type tumors [29]. In a subsequent propensity score matched study, the same investigators showed that the median time to first subsequent therapy (TFST) was significantly longer in the patients who had sCRS in addition to SC and olaparib compared to systemic therapies alone (42 vs. 16 months; p=0.05) and patients undergoing sCRS also had a more prolonged 3-year post recurrence survival (79 vs. 42 %; p=0.02) [30].

Conversely, another retrospective study of 127 patients showed a benefit of sCRS irrespective of the BRCA1/BRCA2 mutation status. However, only 4 % of the patients received maintenance olaparib [31]. More evidence is needed on the role of sCRS in patients with BRCA mutation or homologous repair deficiency, those receiving PARP inhibitors, or bevacizumab as second-line maintenance therapy. Current ESMO-ESGO guidelines still recommend sCRS for oligometastatic, completely resectable, recurrent disease, irrespective of targeted therapies. The role of sCRS in PARP inhibitors (niraparib) is being evaluated in a phase II multi-centre RCT (SGOG SOC-3 Study, NCT03983226).

With more and more patients receiving PARPi during first-line therapy, the role of sCRS in patients with/without BRCA mutations will need careful evaluation [32, 33].

Two recent publications showed that patients having oligometastatic progression on PARP inhibitor maintenance therapy, either during first-line treatment or for recurrence, benefit from locoregional treatment, including surgery [34, 35].

The indications and timing of sCRS for recurrent ovarian cancer are evolving in patients who have received first-line maintenance and will receive second-line maintenance therapy Figure 2.

Figure 2:

Algorithm for selection of patients for secondary cytoreductive surgery and systemic therapies. PFI, platinum-free interval; CGR, complete gross resection; QOL, quality of life; sCRS, secondary cytoreductive surgery; SC, ystemic chemotherapy; Bev, Bevacizumab.

Do all patients undergoing sCRS need second-line SC?

SC is the current standard of care for the treatment of recurrent EOC [36, 37]. sCRS is recommended for selected patients. Could SC be omitted or deescalated in patients with BRCA mutated tumors undergoing sCRS? Future trials should answer this question. Current ESMO-ESGO guidelines and NCCN guidelines do not recommend giving NACT to patients in whom sCRS is planned. There are two situations:

1. The first situation is when the recurrent disease is unresectable, and SC is used to reduce the disease burden and enable resection. sCRS may be of limited benefit since these patients generally have ascites and extensive peritoneal involvement. Intraperitoneal chemotherapy could have a role in these difficult-to-treat patients, which should be investigated in future studies [38].

2. The second situation is when SC is administered first, even when the disease can be removed surgically. SC may allow testing of the disease biology and responsiveness to SC. Complete responders could be treated with consolidation or maintenance therapy, especially if significant morbidity is expected. A schema for selecting patients for sCRS and SC is outlined in Figure 2. A recent randomized trial showed the benefit of adding hyperthermic intraperitoneal chemotherapy to sCRS. Though the entire publication is awaited, it would be interesting to see where HIPEC fits in with the systemic and maintenance therapies [39].