In most of cases, CA 125 elevation is the first signal of recurrence; it can be observed some months before clinical symptoms or radiological signs. It remains difficult in routine practice to know if a treatment has to be immediately initiated. The MRC OV05/EORTC 55955 addressed this question in a randomised trial comparing early versus delayed treatments in case of asymptomatic rising of CA125. No difference was observed in overall survival (OS) in the delayed treatment group (median delayed time: 4.8 months) in comparison with the earlier treated one; furthermore, the quality of life in the delayed group was better [7]. Since the results of that study, there have been a lot of debates on the needs to better document the recurrence (particularly with PET CT), to explore the role of surgery and to evaluate the role of alternative treatments to chemotherapy in such a situation. However, it is commonly assumed that raised CA125 alone is not sufficient to systematically recommend treatment.

Until now, mainly for clinical trial design, relapses have been defined according to time spent between end of the last platinum-based chemotherapy and detection of recurrence, which is called “platinum-free interval (PFI)”. Then relapses are named as platinum refractory (0–3 months), platinum resistant (<6 months), platinum partially sensitive (6–12 months) and platinum sensitive (>12 months) [8]. These definitions are meant to move on with better knowledge on biological behaviour of each tumour. The introduction of non-platinum chemo regimens and targeted therapies may also jeopardise these predefined intervals. They are currently being discussed (5th Ovarian Carcinoma Consensus Conference, Tokyo, November 2015) and are awaiting publication. It seems that we may consider only early and delayed relapses as a reflection of tumour ability to respond to subsequent medical treatments. The delay of relapse is also an important criterion to select patient amenable to surgery.

Surgery at relapse is commonly performed in routine practice based on some retrospective data [9, 10]. It usually concerns first relapses with low volume, accessible anatomical sites and long treatment-free interval. Disease-free interval is the most important prognostic factor at relapse; more than 6 months is generally mandatory to consider secondary cytoreductive surgery (SCS) [11]. The later the relapse, the better the benefit of surgery seems to be; however, the best disease-free interval to indicate surgery was not well established. A meta-analysis with heterogeneous and retrospective studies has shown correlation between residual disease after surgery and overall survival [12, 13]. There are no evidence-based phase III results to validate this routine practice, and the place of SCS remains to be defined yet. The Fourth Ovarian Cancer Consensus Meeting recommendations were that a complete resection must be the main objective of a trial designed to access surgery at relapse [14]. Two prospective trials have addressed this issue, the DESKTOP III trial and GOG 0213, at first relapse beyond 6 months. These two trials compare platinum-based chemotherapy with or without SCS with the aim of no residual disease. Their principal objective is overall survival. Selection criteria for SCS were studied by the Arbeitsgemeinschaft Gynakologische Onkologie (AGO) group in two consecutive studies. The DESKTOP I trial validated a score to identify patients who can benefit from the surgical approach [15]. This score has three items: it is positive if the primary surgery was complete, patient performance status is good and there is no ascites. In every other situation, the score is negative. The DESKTOP II trial prospectively validated the impact of a positive AGO score to obtain a complete resection at relapse [16]. Finally, the DESTKTOP III randomised trial was designed to compare, in case of relapse beyond 6 months and positive AGO score, surgery followed by platinum-based chemotherapy versus platinum-based chemotherapy alone. GOG 0213 trial shared the same main selection criteria. These two prospective trials are closed, and results are awaited in the next years.

They are defined by relapses during first-line treatment or in the few following months. They are called refractory (0–3 months after) or resistant (3–6 months after). But platinum resistance is the final event of all relapses of EOC. As such, they represent a very heterogeneous group of various biological tumour behaviours. They are difficult to treat due to the lack of response to chemotherapy, and for the more previously treated patients, persistence of side effects can be limiting. In these bad prognosis situations, the main objective of treatment is to give the best efficacy, and, as it is uncertain, attention must be focused on toxicity profile. The recommendation in this situation is not to reintroduce platinum especially for refractory relapses. However, in case of platinum-resistant disease, some reports mentioned responses to platinum-based chemotherapy [17, 18]; it is for sure a way to investigate further. Combinations of chemotherapy are not superior to monotherapy and responsible of more side effects. The main available drugs are represented by weekly paclitaxel, liposomal doxorubicin, topotecan and gemcitabine [19–23]. These four drugs offer similar objective responses rates (10–20 %), median progression-free survival (PFS) (3–4 months) and OS (around 12 months) with different toxicity profiles. The AURELIA trial evaluated in a phase III randomised trial the addition of bevacizumab to chemotherapy versus chemotherapy alone in first or second relapses of EOC. Chemotherapy was according to investigator’s choice weekly paclitaxel, topotecan or pegylated liposomal doxorubicin. After completion of chemotherapy, bevacizumab was given in maintenance until progression or toxicity. Better responses (27.3 versus 11.8 %, p = 0.001) and PFS (HR 0.48; 95 % CI 0.38–0.60; p <0.001) were observed with chemotherapy associated with bevacizumab [24]. Median PFS was 6.7 months with chemotherapy plus bevacizumab versus 3.4 months with chemotherapy alone. Responses were of particular interest in cases of ascites with a significant impact in reductions of paracentesis needs. Combination of chemotherapy and bevacizumab was associated with improvement of symptoms as evaluated by patient-reported outcomes [25]. No benefit was observed in OS. Bevacizumab (Avastin™) was licensed in this setting.

Other anti-angiogenic agents (trebananib, pazopanib, nintedanib, cediranib) showed some efficacy in phase II or III studies but none has been licensed yet [26–29]. The phase III trial TRINOVA 1 [26] compared weekly paclitaxel + trebananib (P + T) or placebo (P) in recurrent EOC with PFI <12 months and less than three previous lines of chemotherapy. The results showed a better PFS with weekly paclitaxel associated with trebananib (HR 0.66; 95 % CI 0.57–0.77; p <0.001) 7.2 months versus 5.4 months. They also interestingly displayed a benefit in OS in the subgroup with ascites. In this case, with a median follow-up of 17.7 m, the median OS was 14.5 m (P + T) versus 12.3 m with P alone (HR = 0.72, 95 % CI 0.55–0.93; p = 0.011) [30]. No benefit was observed in OS in the intent to treat population as in previous studies with bevacizumab. The mechanism of action of trebananib is very different from monoclonal vascular endothelial growth factor antibody bevacizumab. It is a peptibody (Fc-peptide fusion molecule) blocking binding of angiopoietins Ang1/2 to the Tie 2 receptor. The toxicity profile is also different with mainly oedema, ascites and pleural effusion and less cardiovascular and renal effects.

They are called platinum sensitive [31, 32] and generally more responsive to chemotherapy. Chemosensitivity is supposed to increase with the widening of the interval. For first relapse, platinum-based chemotherapy is still the most active agents with comparable efficacy of cisplatin and carboplatin [33, 34] but with a toxicity profile in favour of carboplatin. Combination of carboplatin with a second drug (paclitaxel, gemcitabine, pegylated liposomal doxorubicin) has shown better efficacy than platinum monotherapy; it also gives more manageable side effects [32, 35, 36] depending on the second associated drug. A meta-analysis published in 2013, using individual patient data, demonstrated with inclusion of four randomised trials (exploring platinum monotherapy versus platinum combination) and 1300 patients, the superiority of the platinum combination in terms of PFS (HR = 0.68; 95 % CI 0.57–0.81; p < 0.001) and OS (HR = 0.80; 95 % CI 0.64–1.00; p = 0.05) [37]. As patients will be exposed to subsequent lines of chemotherapy, the issue of tolerance must be kept in mind in order to avoid excessive or cumulative toxicities. The CALYPSO trial [36] compared carboplatin plus 3 weekly paclitaxel and carboplatin plus pegylated liposomal doxorubicin. It was designed as a non-inferiority trial. The results showed a better PFS for platinum-pegylated liposomal doxorubicin (HR 0.82, 95 % CI 0.72–0.94) with better toxicity profile and less discontinuing of treatment.

Some have explored an artificial extension of PFI by delivering a non-platinum-based chemotherapy. The OVA 301 trial compared a population of relapsing EOC regardless of PFI in a phase III randomised study pegylated liposomal doxorubicin alone or in combination with a new drug named trabectedin [38]. Trabectedin is a new compound with original antineoplastic properties, the best known being the possibility to bind DNA minor groove. The results showed for platinum-sensitive patients an improved PFS with the combination (HR 0.73; CI 95 % 0.56–0.95; p = 0.0170) and better objective response rate 35.3 % versus 22.6 %. Observed toxicities with the combination were mainly neutropenia, grade ¾ transaminase transient elevations. Meaningful results were observed in the partially platinum-sensitive population (PFI between 6 and 12 months) in which rechallenging with platinum-based chemotherapy could be difficult or not wished. PFS was improved in the combination arm (HR = 0.65 CI 95 % 0.45–0.92) as well as OS (HR = 0.59; CI 95 % 0.43–0.82) [39]. To definitively confirm this concept, a prospective trial was designed, INOVATYON trial, in the partially platinum-sensitive population, comparing platinum-based chemotherapy with PLD-carboplatin with crossover at progression. This study is still ongoing.

More recently, the place of bevacizumab was evaluated in a phase III randomised trial (OCEANS) comparing the doublet carboplatin-gemcitabine with (experimental arm) or without bevacizumab (standard arm) for patients bevacizumab naive [40]. Bevacizumab was given during chemotherapy then in maintenance until progression or unacceptable toxicity. The study met its primary objective, PFS increased in the chemo-bevacizumab arm with 12.4 months versus 8.4 months (HR 0.484; 95 % CI 0.388–0.605; p <0.0001). The objective response rate was also significantly improved: 78.5 versus 57.4 % (p <0.0001). However, no benefit was seen in OS. Treatment was well tolerated with no new signal. No gastrointestinal perforation was observed. The main grade 3 or more toxicities were hypertension (17.4 versus 0.4 %) and proteinuria (8.5 versus 0.9 %). In Europe, these results led the license of bevacizumab (Avastin™) in association with carboplatin-gemcitabine in this setting. However, bevacizumab was already indicated in primary treatment in association with carboplatin-paclitaxel combination in advanced EOC. There is no data concerning benefit of bevacizumab rechallenging in relapsed EOC. That hypothesis is tested in an ongoing Italian phase III trial (MITO 16 study) comparing platinum-based chemotherapy (carbo-paclitaxel, carbo-gemcitabine or carbo-pegylated liposomal doxorubicin) with or without bevacizumab.

Another anti-angiogenic compound has been evaluated in late recurrences. The ICON 6 trial compared in a randomised phase III, cediranib during platinum-based chemotherapy followed by cediranib in maintenance for 18 months versus cediranib during platinum-based chemotherapy followed by placebo in maintenance during the same duration versus platinum-based chemotherapy with placebo. The results demonstrated a better PFS (HR 0.56; 95 % CI 0.44–0.72; p < 0.0001), OS data are immature yet (HR 0.77; 95 % CI 0.55–1.07; p = 0.11) in arm with cediranib during chemotherapy and in maintenance in comparison to the standard arm with chemotherapy alone [41]. The main side effects were hypertension, fatigue, nausea and diarrhoea.

Cediranib is a small oral molecule VEGF receptor tyrosine kinase inhibitor, which also inhibits both platelet-derived and fibroblast growth factor receptors, explaining the different toxicity profile.

This new compound was recently assessed in a randomised phase II in association with olaparib versus olaparib alone in 86 patients with a platinum-sensitive ovarian cancer relapse. This “all oral no-chemo doublet” showed impressive results in term of efficacy: PFS was extended by 8.7 m (9 m with olaparib alone versus 17.7 m with the combination). The improvement was observed regardless of germinal BRCA status. Toxicity was significant with mainly grade 3/4 hypertension, fatigue and diarrhoea [42]. The place of such combination remains to be evaluated on larger population. Olaparib belongs to PARP inhibitors family and is so far the most evaluated agent. Olaparib (Lynparza™) was licensed in Europe, in platinum-sensitive relapse of m BRCA serous ovarian cancer, in case of response to platinum-based chemotherapy, following the results of Study 19 [43]. This phase II randomised study compared, after good response to platinum-based chemotherapy, treatment by olaparib 400 mg twice a day (introduced after the chemotherapy) versus placebo in 265 women with platinum-sensitive relapse: adding olaparib showed a benefit. In that study, 51 % of patients had germinal or somatic BRCA mutations. The benefit of olaparib was greater in the mBRCA group with a reduced risk of progression of 82 % (HR 0.18; 95 % CI 0.11–0.31, p <0.00001). The median PFS was 11.2 m in the mBRCA group versus 4.3 m for the others. However, data for OS are not mature yet.