Het pancreascarcinoom wordt gezien als een systemische ziekte, niet alleen gebaseerd op het hoge percentage patiënten dat zich met gemetastaseerde ziekte presenteert, maar tevens op basis van het hoge percentage snelle recidieven na (radicale) resectie. Derhalve is (adjuvante) chemotherapie een belangrijk onderdeel van de behandeling van het pancreascarcinoom. Het naar voren halen van de systemische behandeling naar de neo-adjuvante setting zal de komende periode worden geëvalueerd. Vooralsnog is de standaard behandeling een resectie, gevolgd door adjuvante chemotherapie.

Description of studies

Adjuvant treatments compared to observation only

Adjuvant treatments compared to observation only were studied in several trials: ESPAC-1 (Neoptolemos 2001; Neoptolemos 2004; a total of 545 patients among 3 subgroups), ESPAC-1+ (Neoptolemos 2009a), ESPAC-3v1 (Neoptolemos 2009b, n=314 in ESPAC-1+ and ESPAC-3v1 combined), CONKO-001 (Oettle 2007; Oettle 2013, n=354), JSAP-02 (Ueno 2009; n=118) and the earlier trial by JSAP (Kosuge 2006; n=89), , and EORTC 40891 (Smeenk 2007; n=120) and the trial by Kalser (1985) (detailed description of studies in Liao 2013) (Table 1).

After Oettle (2013) no studies comparing adjuvant therapy to surgery alone were published. Oettle (2013) described the long-term follow-up results of the CONKO-001 trial, in which adjuvant treatment with gemcitabine was compared with observation only. With adjuvant therapy as the standard of care (in the Netherlands: GEM), the opportunity to compare adjuvant treatment against observation alone, within the context of a randomized trial format, may not be repeated.

Hazard ratios (HRs) were explicitly reported in five studies and could be estimated by Liao (2013) in three studies. One study (Kalser 1985) reported median survival durations. The ESPAC-1 trial included three subgroups; for the subgroup with two-by-two factorial design Liao (2013) used the overall survival updated in a later report (Neoptolemos 2004) in the analysis. One publication (Neoptolemos 2009) reported composite data of ESPAC-1, ESPAC-1+, and ESPAC-3 v1, thus Liao (2013) included only data from ESPAC-1+ and ESPAC-3 v1 to avoid duplication. We extracted HRs from Liao (2013) and used the results of long-term follow-up of Oettle (2013) instead of their earlier results (Oettle 2007). For toxic effects, Liao (2013) compared the five adjuvant treatments and calculated odds ratios (ORs) from the number of total patients and the number of patients with toxic effects in each trial for meta-analysis.

Table 1 Details of the studies comparing adjuvant treatments and observation only

Adjuvant chemotherapy compared to adjuvant gemcitabine chemotherapy

Xu (2017) compared in a network meta-analysis all adjuvant chemotherapy treatments: a total of nine chemotherapy regimens including S-1, fluorouracil (5FU), gemcitabine(gemcitabine), gemcitabine plus capecitabine, 5FU plus chemoradiation (CRT), gemcitabine plus CRT, cisplatin plus epirubicin plus 5FU plus gemcitabine plus CRT, gemcitabine plus uracil/tegafur and 5FU plus cisplatin were involved (Xu 2017). Xu (2017) reported five trials in which the control group received gemcitabine treatment: ESPAC-4 (Neoptolemos 2017), JASPAC01 (Uesaka 2016; Hagiwara 2018), Shimoda (2015), ESPAC-3v2 (Neoptolemos 2010), and Yoshitomi (2008). An overview of the studies relevant for answering the PICO on adjuvant chemotherapy is presented in table 2.

Table 2. Details of the studies comparing adjuvant chemotherapy and adjuvant treatment with gemcitabine

Results

1.1 Adjuvant chemotherapy compared to observation only

1.1.1 Overall survival (critical outcome)

Both adjuvant chemotherapy with 5FU or gemcitabine decreased overall mortality (see Figure 1) compared to observation only after surgery. Pooled results for both adjuvant chemotherapy treatments showed an HR of 0.68 (95% CI 0.57 to 0.81), indicating that adjuvant chemotherapy decreased the risk of mortality after surgery by about a third. Kosuge (2006) evaluated the effect of adjuvant cisplatin and 5FU therapy after curative resection among a total of 89 patients. They found that median survival was 15.8 months in the surgery-alone group and 12.5 months in the surgery + chemotherapy group, and the 5-year survival rate was 14.9% in the surgery-alone group and 26.4% in the surgery + chemotherapy group (p = 0.94).

Figure 1. Pooled hazard ratios for death by traditional meta-analysis (adjuvant chemotherapy versus observation).

1.1.2 Relapse-free survival

Available data on relapse-free survival are presented in Table 3. The median time to relapse ranges from 10.2 to 13.4 months for patients receiving adjuvant chemotherapy. For patients receiving only surgery, the median time to relapse ranges from 5 to 8.6 months.

Table 3. Relapse free survival (median time, 95% CI) (adjuvant chemotherapy versus observation)

1.1.3 Toxicity

A total of 70 patients receiving adjuvant treatment with 5FU or gemcitabine (of a total of 334 patients receiving adjuvant treatment in 4 studies, 21%) experienced grade 3-4 toxic effects. Liao (2013) calculated an estimated odds of 0.33 in favour of 5FU compared to gemcitabine, but this difference in the proportion of patients experiencing toxic effects was not statistically significant different (OR 0.33; 95% CI 0.03 to 4.29).

1.1.4 Quality of Life (critical outcome)

Carter (2009) calculated Quality adjusted life months (QALM) over 24 months. For the chemotherapy group, the QALM-24 was 9.6 (95% CI: 8.7 to 11.2) months compared with the mean QALM-24 of 8.6 (95% CI: 7.6, 10.5) months for the no chemotherapy group.

Rating of the overall quality of evidence for each outcome

The quality of the evidence comparing adjuvant chemotherapy to observation only:

• was ‘High’ for the outcomes ‘overall survival’ and ‘relapse free survival’;
• was lowered from ‘High’ to ‘Low’ for the outcome ‘toxicity’. Reasons for downgrading the body of evidence were: clinical heterogeneity, imprecision (the confidence interval crossing 1) and indirectness (no direct comparisons for all adjuvant treatments with observation only);
• was lowered from ‘High’ to ‘Low’ for the outcome ‘quality of life’. Reasons for downgrading the body of evidence were some methodological limitations (incomplete QoL data accrual and fallout through death) and serious imprecision (results from one study and a (relatively) small number of patients randomized).

There was no indication of publication bias.

1.2. Adjuvant chemotherapy versus adjuvant treatment with gemcitabine

1.2.1 Overall survival (critical outcome) (Table 4)

One large RCT (n=377) (Uesaka 2016) suggested that S-1 adjuvant chemotherapy prolonged the median survival (46.5 months, 95% CI 37.8 to 63.7) in Asian patients compared to gemcitabine adjuvant therapy (25.5 months, 95% CI 22.5 to 29.6; HR 0.57, 95% CI 0.44 to 0.72). One smaller RCT including 57 patients (Shimoda 2015) did not find a beneficial effect of S-1 treatment compared to gemcitabine in terms of overall survival (HR 0.70, 95% CI 0.36 to 1.36, p=0.293).

Gemcitabine plus capecitabine also showed a favourable effect on mortality (HR=0.82, 95% CI 0.68 to 0.98) (Neoptolomos 2017).

Overall survival after treatment with gemcitabine plus uracil/tegafur or fluorouraciI alone was not statistically significant different compared to gemcitabine alone (Neoptolemos 2010; Yoshitomi 2008). Furthermore, the network meta-analysis of Xu (2017) showed that chemotherapy plus 5FU or gemcitabine is inferior to 5FU or gemcitabine alone in terms of overall survival.

Table 4. Overall survival (median time, 95% CI) comparing adjuvant chemotherapy with adjuvant treatment with gemcitabine.

* Estimated data, awaiting long term results

1.2.2 Relapse-free survival

Uesaka (2016) reported a median relapse free survival time of 22.9 months (95% CI 17.4 to 30.6) after S-1 adjuvant treatment compared to 11.3 months (9.7 to 13.6) after adjuvant gemcitabine treatment. Shimoda (2015) found that after S-1 adjuvant treatment relapse-free survival tended to be longer compared to gemcitabine, but the 4-month difference was not statistically significant (Table 3). Other adjuvant treatments did not result in a longer relapse-free survival time.

Table 5. Relapse free survival (median time, 95% CI) (adjuvant chemotherapy versus adjuvant treatment with gemcitabine)

1.2.3 Toxicity

The network meta-analysis of Xu (2017) showed that S-1 chemotherapy ranked the least toxic, whereas gemcitabine plus capecitabine ranked the most toxic. 5FU showed more toxic events than in the group of gemcitabine. In table 4 the number of patients experiencing toxic effects are shown for each adjuvant chemotherapy.

Table 6. Toxicity (number of patients experiencing toxic effects (adjuvant chemotherapy versus adjuvant treatment with gemcitabine)

1.2.4 Quality of Life (critical outcome)

Neoptolemos (2010) reported quality of life (QoL) within a 12-month duration from surgery for patients receiving 5FU compared with those receiving gemcitabine. They found no statistically significant differences in mean standardized area under the curve scores for global quality-of-life scores across treatment groups conditional on patient survival; mean standardized AUC was 43.6 (SD 20.1) for patients receiving 5FU (plus folinic acid), compared with 46.6 (SD 19.7) for those receiving gemcitabine (p=0.08).

Hagiwara (2018) reported, based on per-protocol analyses, that health-related QoL within 6 months from treatment initiation was comparable between the patients receiving S-1 and gemcitabine, but it was better thereafter in the S-1 group. QALM-24 and Quality adjusted relapse free months (QARFM-24), measures that simultaneously take survival outcomes and QOL into account, were both longer in the S-1 group (17.3, 95% CI 16.5 to 18.2 and 14.6, 95% CI 13.6 to 15.7, respectively) than in the gemcitabine group (14.8, 95% CI 13.9 to 15.7 and 11.3, 95% CI 10.3 to 12.3, respectively). The difference in QALM-24 was 2.5 months in favour of S-1 (95% CI 1.3 to 3.8, p < 0.001) and in QARFM-24 the difference was 3.3 months in favour of S-1 (95% CI 1.9 to 4.8, p < 0.001).

Rating of the overall quality of evidence for each outcome

The quality of the evidence comparing adjuvant chemotherapy to adjuvant chemotherapy with gemcitabine:

• was lowered from ‘High’ to ‘Low’ for the outcomes ‘overall survival’ ‘and ‘relapse free survival’ for all other adjuvant treatments compared to gemcitabine. Reasons for downgrading the body of evidence were: clinical heterogeneity (regarding S1, with one study in an Asian population showing an inexplicable large effect, and another study showing no effect), and serious imprecision (regarding other adjuvant treatments with results originating from single studies).
• was lowered from ‘High’ to ‘Moderate’ for the outcome ‘toxicity’ due to clinical heterogeneity and some studies showing more toxic events than gemcitabine and others less events.
• was lowered from ‘High’ to ‘Low’ for the outcome ‘quality of life’. Reasons for downgrading the body of evidence were some methodological limitations (incomplete QoL data accrual and fallout through death) and clinical heterogeneity.

There was no indication of publication bias.

2. Adjuvant chemoradiation compared to observation only

2.1 Overall survival (critical outcome)

In three publications the effect of 5FU-based chemoradiation after surgery compared to no adjuvant treatment on mortality was reported. Estimated effects were both negative and positive. Pooled results showed no survival advantage of adjuvant chemoradiation therapy (HR 1.01; 95% CI: 0.79 to 1.29).

Figure 2. Pooled hazard ratios for death by traditional meta-analysis (adjuvant 5FU-based chemoradiation versus observation).

2.2 Relapse-free survival

Smeenk (2007) reported a median relapse free survival period of 18 months (95% CI 12 to 21.6) in the intervention group versus 14.4 months (95% CI 10.8 to 20.4) in the observation only group.

2.3 Toxicity

A total of 12 patients receiving adjuvant chemoradiation (of a total of 136 patients receiving this adjuvant treatment in 2 studies, 9%) experienced grade 3-4 toxic effects. Liao (2013) calculated that chemoradiation was ranked as the least toxic adjuvant treatment.

2.4 Quality of Life (critical outcome)

Carter (2009) studied a subset of ESPAC-1 patients (n=316) and calculated for the chemoradiation group a mean QALM-24 of 7.1 (95% CI: 6.0 to 9.0) months compared with a mean QALM-24 of 8.1 (95% CI: 7.0 to 10.0) months for the no chemoradiation group.

Rating of the overall quality of evidence for each outcome

The quality of the evidence comparing adjuvant chemoradiation to observation only:

• was lowered from ‘High’ to ‘Moderate’ for the outcome ‘overall survival’. The risk of bias due to methodological constraints was considered to be low (Xu 2017; Liao 2013). However, the quality of the evidence was downgraded one level due to some unexplained statistical heterogeneity (I²=64%).
• was lowered from ‘High’ to ‘Low’ for the outcome ‘relapse-free survival’ due to serious imprecision (results from one study and hence, a small sample size).
• was lowered from ‘High’ to ‘Low’ for the outcome ‘toxicity’ due to serious imprecision (few patients involved).
• was lowered from ‘High’ to ‘Low’ for the outcome ‘quality of life’. Reasons for downgrading the body of evidence were some methodological limitations (incomplete QoL data accrual and fallout through death) and serious imprecision (results from one study).

3. Adjuvant chemoradiation plus chemotherapy compared to observation only

3.1 Overall survival (critical outcome)

One study assessed the effect of chemoradiation plus 5FU versus observation. No studies compared chemoradiation plus gemcitabine with observation only.

In 1985, the Gastrointestinal Tumor Study Group (GITSG) randomized a total of 43 patients to surgery alone or additional chemoradiotherapy followed by chemotherapy. The median survival was significantly longer in the adjuvant treatment group (20 months vs 11 months), with resp. 43% and 18% survival after 2 years and 18% and 8% after 5 years (Kalser 1985). The study was terminated after the inclusion of only 43 patients due to poor accrual and it was unclear whether the improved survival was due to chemoradiotherapy or chemotherapy alone.

Liao (2013) presented the results of studies comparing chemoradiation plus 5FU or gemcitabine with other adjuvant treatments. Liao (2013) found no survival advantage over chemoradiation plus 5FU compared to chemoradiation alone (HR 0.59; 95% CI 0.19 to 1.74) or chemotherapy with 5FU alone (HR 0.87; 95% CI 0.27 to 2.69). No effect (HR 0.82; 95% CI 0.40-1.71) for chemoradiation plus gemcitabine compared to chemoradiation plus 5FU was found either.

3.2 Relapse-free survival

Not reported.

3.3 Toxicity

In the GITSG study a total of 3 patients (3/22; 14%) in the intervention group experienced toxic effects. Liao (2013) calculated (based on mutual comparisons) that chemoradiation plus gemcitabine or 5FU was most likely to be ranked the worst or second worst in terms of overall toxic effects, with estimated ORs ranging from 1.61 (when compared to chemotherapy with 5FU) to 27.80 (when compared to chemoradiation).

3.4 Quality of Life

No RCTs reporting on quality of life measures for patients receiving adjuvant chemoradiation plus chemotherapy were found.

Rating of the overall quality of evidence for each outcome

The quality of the evidence comparing adjuvant chemoradiation plus chemotherapy compared to observation only:

• was lowered from ‘High’ to ‘Very low’ for the outcome ‘overall survival’ due to methodological constraints (terminated inclusion), serious imprecision and indirectness (only one trial assessing the direct effect and few patients involved).
• was lowered from ‘High’ to ‘Low’ for the outcome ‘toxicity’ due to serious imprecision (few patients involved) and indirect comparisons.
• No evidence was found reporting the outcomes ‘relapse free survival’ and ‘quality of life’.

In order to answer the clinical question (‘uitgangsvraag’), we performed a systematic literature search for the following research question(s):

Is adjuvant therapy associated with better survival and quality of life compared to no adjuvant therapy?

P Patients with pancreatic cancer

I Adjuvant radiotherapy, adjuvant chemotherapy, or both

C Surgery only or current care-as-usual, i.e. gemcitabine

O Survival, toxicity (number of patients experiencing toxic effects), relapse-free survival, quality of life (QoL)

Relevant outcome measures

The working group decided that survival and quality of life were crucial outcome measure for decision-making and toxicity, relapse-free survival as important outcome measures for decision-making. The working group did not define the mentioned outcome measures beforehand, however, they used the definitions described in the studies.

Zoeken en selecteren (Methode)

The information specialist searched the databases Medline (via OVID), Embase (via Embase.com) and the Cochrane Library (via Wiley)] on 23rd November 2017, relevant search terms were used to search for English-language systematic reviews (SRs) on adjuvant therapy in patients with pancreatic cancer published from 2010 The search acknowledgement can be found in the tab Acknowledgement.

This literature search yielded 212 hits. Studies were selected on the basis of the following selection criteria:

• Study population concerns patients with pancreatic cancer;
• The intervention is adjuvant radiotherapy, adjuvant chemotherapy or both;
• Concerns systematic review.

On the basis of title and abstract, 15 articles were initially pre-selected. After consulting the full text, 13 studies were then excluded (see exclusion table in the tab Acknowledgement), and 2 SRs were finally selected: Xu (2017) and Liao (2013). Xu (2017) was the most recent SR (search till November 2016) and it was decided to search for randomized controlled trials on the topic published in 2016 or later.

The information specialist searched the databases Medline (via OVID) and Embase (via Embase.com) on 30^th April 2018, relevant search terms were used to search for English-language RCTs on adjuvant therapy in patients with pancreatic cancer. Details of the search strategy can be found in the tab Acknowledgement. This literature search yielded 212 studies published from 2016 onwards. Studies were selected on the basis of the following selection criteria:

• Study population concerns patients with pancreatic cancer;
• The intervention is adjuvant radiotherapy, adjuvant chemotherapy or both;
• Concerns primary (original) comparative research (RCT).

On the basis of title and abstract, two articles were initially pre-selected. After consulting the full text, one study was finally selected: Hagiwara (2018). The other selected study was already included in the SR of Xu (2017).

The most important study characteristics and results are shown in the evidence tables. The judgement of the individual study quality (risk of bias) is shown in the risk of bias tables.

The systematic review of Liao (2013) provided a more comprehensive description of all trials and articles and data were extracted from this review. Data were supplemented with the more recent publication of Oettle (2013). Xu (2017) and Liao (2013) reported overall survival, relapse free survival and toxicity. Therefore, the literature was checked for publications describing the outcome ‘quality of life’ that originated from the trials described in Xu (2017).