Neoadjuvant therapy for downstaging of locally advanced hilar cholangiocarcinoma: a systematic review

Abstract

Background

Hilar cholangiocarcinoma is a rare but highly lethal type of cancer. A minority of patients present with resectable disease. Surgery remains the only treatment modality offering a chance of long-term survival. Unresectable patients are typically offered palliative treatment. The aim of this systematic review was to summarize the evidence for neoadjuvant therapy followed by surgical resection in patients presenting with hilar cholangiocarcinoma.

Methods

Cochrane databases, Medline, PubMed and EMBASE were systematically searched to identify articles describing neoadjuvant therapy and surgical resection or re-assessment of resectability in patients with hilar cholangiocarcinoma. Included were all articles with original research. Study selection and data extraction were performed separately by two reviewers using a standardized protocol.

Results

From 732 articles 8 full text articles and 2 abstracts met the inclusion criteria. The 2 abstracts and 1 full text article were case reports, 3 articles were retrospective and 4 were prospective studies (2 phase I and 2 phase II studies). Photodynamic therapy, chemotherapy and radiation therapy were used in various indications in populations that included patients with hilar cholangiocarcinoma, some of which were primarily unresectable. Overall quality of articles was limited.

Conclusion

Current evidence suggests that neoadjuvant therapy in patients with unresectable hilar cholangiocarcinoma can be performed safely and in a selected group of patients can lead to subsequent surgical R0 resection. Surgical resection of downstaged patients should be assessed in properly designed phase II studies.

Background

Cholangiocarcinomas account for approximately 3% of all gastrointestinal malignancies,^1,2 this represents 5000 new patients in the US annually.^2,3 From an anatomical point of view approximately 60–70% are located within 2 cm from the bifurcation of the common bile duct (hilar cholangiocarcinoma, also called Klatskin tumour), extrahepatic cholangiocarcinomas occur in approximately 20–30% of patients and intrahepatic in 5–10%.^2,4 Surgery represents the only chance of long-term survival and cure and the 5-year survival after an R0 resection of hilar cholangiocarcinoma is in the range of 30–50%.^2,5–11

Unfortunately, the majority of patients who present with hilar cholangiocarcinoma are not resectable owing to the local extent (involvement of vessels or bilateral extension beyond the secondary radicals) or because of the presence of metastatic disease.^2,12,13 This was described in a retrospective review by Jarnagin et al.,⁸ when 65% of the 225 patients with hilar cholangiocarcinoma treated at MSKCC between 1991 and 2001 had unresectable disease, 32% as a result of metastases and 23% because of locally unresectable disease. Patients with unresectable disease are usually referred only for palliative therapy. It may prolong their life as well as control their symptoms with the use of chemotherapy, radiation or photodynamic therapy (PDT).^14–20 The expected survival in this patient population is 6–12 months depending on the therapy used.¹⁶

The outcomes of incomplete R1/R2 resection in spite of the use of adjuvant therapy are comparable with palliative PDT without resection.²¹ This makes decision making in marginally resectable patients challenging. The benefit of an aggressive approach if a R0 resection is achieved needs to be weighed against the impact on quality of life without a survival benefit in patients with a R1/R2 resection.

There is no general consensus when it comes to use of neoadjuvant and adjuvant therapy. Multiple guidelines, reviews and opinions describe that these therapies are currently not indicated. They suggest further research as there is lack of sufficient evidence from prospective studies. The only exception is the use of neoadjuvant chemoradiation prior to the liver transplant in a highly selected group of patients with hilar cholangiocarcinomas. It has been extensively studied and neoadjuvant therapy in this setting is currently a requirement.^22–27

The authors hypothesize that a certain number of patients with locally advanced hilar cholangiocarcinoma, currently only offered palliative therapy, could be rendered resectable with the use of neoadjuvant therapy. Downstaging and subsequent resection could potentially improve their outcome. Similarly, marginally resectable patients might benefit from downstaging of their tumours by increasing their chance for an R0 resection.

The aim of this review is not to compare regimens or to identify the one with the best pathological response. This has been done in the past in great detail¹⁶ and with new emerging chemotherapeutics it will continue to evolve. The aim is to describe clinical relevance of neoadjuvant therapy. Only complete pathological response guarantees clinical significance in terms of subsequent resectability.

Methods

This study was conducted in accordance with the standards of quality for reporting systematic reviews (PRISMA).²⁸ A formal protocol has not been created on the web for access. The authors did not receive any funding for the systematic review.

The authors sought to answer a question: What is the evidence with regards to the safety and efficacy of downstaging locally advanced hilar cholangiocarcinoma with the use of neoadjuvant therapy and subsequently resecting them?

Search strategy

Using the Cochrane databases, Medline, PubMed and EMBASE a search was performed in October 2012 using either keyword or subject heading (MeSH) ‘cholangiocarcinoma’ OR ‘Klatskin tumor’ AND keyword ‘neoadjuvant’ or subject heading ‘neoadjuvant therapy’. Searches were auto exploded and no limits were used. In terms of identification of additional papers during full text reviews, six additional papers were identified and all were added to the full text reviews. The authors kept the initial search terms broad in order to identify all potentially relevant articles.

Study selection

It was very likely that multiple different types of cholangiocarcinomas or even other types of biliary tract cancers were grouped together in some studies. The focus on exclusion of articles that did not contain any patients with specifically hilar cholangiocarcinoma was left for the second stage of the review process. The aim was to include all articles that described at least a subset of patients with hilar cholangiocarcinoma.

Similarly, in spite of a very specific research question in terms of the indication for neoadjuvant therapy, all articles describing safety and efficacy of neoadjuvant therapy followed by resection were included. This would include articles describing the use of neoadjuvant therapy in resectable patients with the aim of improving local control or survival. The rationale for that was a reasonable expectation that resection after neoadjuvant therapy will have a similar morbidity and mortality profile regardless of indication.

As a result, all articles that contained original research regarding any neoadjuvant therapy used for patients with resectable or unresectable hilar cholangiocarcinoma who subsequently underwent either an assessment of resectability or a resection were included. (Fig. 1)

Figure 1.

Flow chart of steps in systematic review

Excluded were all review and expert opinion articles without original data, chemotherapy, radiation or photodynamic therapy in an adjuvant or palliative setting as well as papers describing neoadjuvant therapy followed by an orthotopic liver transplant. Also studies describing only response rates without an assessment of subsequent resectability were excluded because resectability cannot be assumed purely based on overall decreased volume of the tumour.

Two authors (J.G. and P.G.) independently screened titles and abstracts of all studies for inclusion eligibility. Any disagreements were settled by including all of the identified papers in full-text review.

All articles were written either in English or German language, translation from German was performed by the same two authors.

Data extraction

The same two authors then extracted data from included articles; this was again done through independent work. The developed protocol included description of the study method, the number of patients, the reason for unresectability, the type of neoadjuvant regimen, the type of subsequent reassessment of resectability, timing and type of subsequent surgery, outcome after surgery including complications and survival, and the type of conclusion the study suggested.

Results

Trial flow

The search and the whole screening process are described in the flow chart (Fig. 1). Agreement on selected studies was 100%.

Data extraction

Table 1 shows original data extracted from the articles.

Table 1.

Characteristics of included articles and summary of outcomes

	Study	Type of study	Type of neoadjuvant Tx	Indication	N neo	Resectability	Anatomy	Results	Conclusion
1	McMasters et al., 199736	non-randomized prospective study	5-FU and EBRT	Downstaging	9	6 unresectable	5 hilar, 4 distal CC	100% R0 resection	neoadjuvant therapy is safe and improves resectability
2	Berr et al., 200029	case report	PDT	Unknown	1	resectable	Hilar CC	4 mm deep complete response	PDT is feasible as neoadjuvant therapy
3	Gerhards et al., 200034	retrospective review	EBRT	prevention of implantation metastases post ERCP/PTC	19	resectable	Hilar CC	0% implantation metastases versus 20% if no neoadjuvant tx	reduced risk of implantation metastases
4	Wiedmann et al., 200338	non-randomized prospective study	PDT	feasibility of PDT in neoadjuvant setting	7	advanced tumors	Hilar CC	100% R0 resection	low risk therapy with improved resectability
5	Witzigmann et al., 200621	non-randomized prospective study	PDT	only comparison of all resected versus all palliative	8	not described	Hilar CC	87.5% R0 resection	R1/R2 resection not different than palliative PDT (7 patients used also in prev. study)
6	Nelson et al., 200833	retrospective review	5-FU and EBRT ± brachytherapy	Downstaging	12	10 unresectable	Hilar and distal CC	11/12 had R0 resection	similar survival, similar complication rates, improved resectability with neoadjuvant
7	Katayose et al., 201137	phase I study	gemcitabine and EBRT	recommended dose identification	12	resectable	all extrahepatic CCs	RD gemcitabine 600 mg/m2	RD gemcitabine 600 mg/m2
8	Glazer et al., 201232	retrospective review	gemcitabine based, no further specification	comparison of neoadjuvant versus surgery upfront	28	resectable	all biliary tract cancers	surgery delayed by 6.8 months, median survival 42.3 versus 53.5 months if no neoadjuvant	do not delay surgery with neoadjuvant therapy if resectable
A1	Sano et al., 201131	case report	Gemcitabine	Downstaging	1	unresectable	Hilar CC	18 months post op no recurrence	potential therapeutic option
A2	Tada et al., 201230	case report	gemcitabine and S-1	Downstaging	1	unresectable	Hilar CC	29 months post op no recurrence	potential therapeutic option

n neo, sample size receiving neoadjuvant therapy, EBRT, external beam radiation therapy, CC, cholangiocarcinoma, PDT, photodynamic therapy.

Case reports

The authors identified three case reports;^29–31 two of them^30,31 were added to the overview only based on abstracts as full text articles were not available for review. These two abstracts describe two patients with unresectable hilar cholangiocarcinoma who successfully underwent a surgical resection after being downstaged with neoadjuvant therapy. Tada et al.³⁰ performed an extended left hepatectomy with partial resection of the portal vein, Sato et al.³¹ describe an extended left hepatectomy with a caudate lobectomy with resection and reconstruction of the hepatic artery. In both cases all margins were negative. With all limitations, these case reports show at least a potential for aggressively treating a selected group of patients who would only be offered palliative treatment based on current recommendations.

The third case report²⁹ describes neoadjuvant photodynamic therapy used in a resectable patient with hilar cholangiocarcinoma who underwent a resection of extrahepatic bile duct together with a Whipple's procedure on the 23^rd day after PDT. Histological examination revealed a complete response to the depth of 4 mm and this paper suggests that PDT is feasible as a neoadjuvant therapy. It is important to stress the fact that this case suggests the use of neoadjuvant PDT in resectable patients. The indication for that is not clearly stated in the study. We can speculate that it was likely to improve local control as opposed to downstaging which would be difficult to achieve with the result PTD offers.

Retrospective studies

Retrospective review papers^32–34 used hospital databases to describe and compare neoadjuvant and adjuvant therapy.

The first one, by Nelson et al.,³³ identified a cohort of 12 patients out of 45 with both proximal and distal extrahepatic cholangiocarcinomas who underwent neoadjuvant therapy. 10 were initially unresectable and 2 were resectable but treated with external beam radiation therapy and fluoropyrimidine-based chemotherapy as sensitizer because of physician preference. The other 33 patients underwent resection followed by adjuvant therapy. As a result, 11 out of the 12 patients in neoadjuvant group underwent an R0 resection, 3 with complete pathological response. There was similar surgical morbidity and 5-year survival of 53% in the neoadjuvant group versus 23% in the adjuvant group which was statistically non-significant.

The second study by Glazer et al.³² grouped all 157 resectable patients with gallbladder cancer and cholangiocarcinoma and compared neoadjuvant, adjuvant chemotherapy and adjuvant chemoradiation cohorts in their outcomes. The study suggested that neoadjuvant therapy in 28 patients delayed surgery on average by 6.8 months and with that resulted in a median survival of 42.3 versus the 53.5 months for the upfront resected cohort, which was statistically significant. Unfortunately, no patient characteristics or selection criteria and indications were described, besides the fact that a panel of experts decided which patient was going to be treated using which strategy. It is reasonable to therefore hypothesize that patient selection bias is a major factor influencing outcome. The neoadjuvant group then possibly included patients with AN overall worse prognosis.

The third study by Gerhards et al.³⁴ describes the use of neoadjuvant external beam radiation therapy using 3 × 3.5 Gy as a means of decreasing the risk of implantation metastases after ERCP or PTC drainage procedures. With this strategy none of the 19 patients with preoperative drainage procedures and neoadjuvant EBRT developed implantation metastases versus 20 % in case of preoperative drainage and no neoadjuvant therapy described in a previous study.³⁵

Prospective studies

Only four prospective studies that fulfilled the inclusion criteria were conducted. Two were phase I studies^36,37 and two were phase II studies.^21,38

McMasters et al.³⁶ in the original article ‘Neoadjuvant Chemoradiation for Extrahepatic Cholangiocarcinoma’ in 1997 suggested the idea of pre-operative neoadjuvant chemoradiation with the intention to downstage hilar and distal cholangiocarcinomas and improve R0 resectability. In their non-randomized study they treated a total of nine patients, six of which were deemed unresectable, with a combination of 5-FU chemotherapy and radiation therapy without a standardized protocol (neoadjuvant external beam radiation therapy in doses ranging from 30 to 50.4 Gy, post-operative 10 Gy brachytherapy and three distal cholangiocarcinomas received 10 Gy intra-operative external beam radiation boost). As a result, all nine patients underwent an R0 resection, compared with 45% (14/31) in a group that did not receive any neoadjuvant therapy. A complete response occurred in three patients. There were no major complications in the neoadjuvant group (three minor wound complications and one cardiac arrhythmia). As a result, this study suggested that neoadjuvant chemoradiation therapy can be done safely, improves the ability to perform a resection with tumour-free margins and provides a significant antitumour response.

The limitations of this study are related to different doses and timing of neoadjuvant therapy for each patient as well as unclear selection of patients with hilar and distal cholangiocarcinoma, both resectable and unresectable. Nevertheless, this study does describe patients with initially unresectable hilar cholangiocarcinoma who were able to safely undergo an R0 resection.

The second phase I study is a Japanese NACRAC study by Katayose et al.,³⁷ which suggested that the recommended dose of gemcitabine when combined with radiation as neoadjuvant therapy is 600 mg/m². This study gave the foundation for an ongoing phase II study, interim results of which were presented at the 2011 European Multidisciplinary Cancer Congress in Stockholm, Sweden in September 2011 as a poster.³⁹ These results suggested safety of phase II so far and continuation of this study. No further details are available to date.

The other two prospective studies are both phase II non-randomized trials. In 2003, Wiedmann et al.³⁸ published a single arm trial in which seven patients with advanced hilar cholangiocarcinoma were treated with neoadjuvant photodynamic therapy and subsequently underwent a surgical resection. Initial resectability is not described, the paper only describes ‘advanced hilar cholangiocarcinoma’ which were Bismuth–Corlette types III and IV and type II with wide extension into the bile duct. All patients were subsequently resected with clear margin and minimal complications. Based on this study it is not possible to determine what the clinical implication of this result is besides the suggestion that it is a low-risk therapy that results in complete tumour destruction into the depth of 4 mm.

The second phase II study²¹ uses the updated Wiedmann's³⁸ database of patients. The main objective of this article WAS the comparison of curative intent surgical and palliative patients in their outcomes. A subset of eight patients after neoadjuvant therapy is a part of the surgical arm. Again, the indication was ‘advanced hilar cholangiocarcinoma’ without description of resectability. This study has not focused on a description of the neoadjuvant group separately in all outcomes. As a result the conclusion can only be that in this cohort seven out of eight patients after neoadjuvant therapy underwent an R0 resection and that there was a trend towards improved survival with a 5-year survival of 42 versus 19% in the upfront surgery group. The major limitation of these two trials^21,38 is the possibility that all eight patients could have undergone an R0 resection without the use of PDT. There is no evidence that PDT improves resectability or helps with downstaging. The conclusion should in that case be only the safety of resection after PDT.

Discussion

Literature extensively describes the pathological response of unresectable hilar cholangiocarcinoma to chemo, radiation and photodynamic therapy in a palliative setting.^14–20 It is not well codified how this translates to a clinical response in terms of subsequent resectability and outcomes after a resection.

During the systematic review of the current literature, the authors identified several hundred guidelines, non-systematic reviews and expert opinions on this topic. The vast majority of them suggested that there is lack of evidence to support the use of neoadjuvant therapy. They suggest aggressive upfront surgical therapy as the only hope for a cure or long term survival. They also suggest accepting the significant risk of performing R1 or R2 resections which have outcomes comparable to palliative therapy only. It is both surprising and concerning that these recommendations are made based on maximum eight studies and two case reports that actually do not present any evidence to suggest against such strategy at all. It is true, that their poor quality, low sample sizes and bias associated with case studies, retrospective and non-randomized prospective studies does not offer sufficient evidence to support neoadjuvant therapy and re-assessment for a resection. On the other hand, the results of these studies are encouraging enough to suggest development of properly designed phase II studies in order to offer answers to the research question.

Most common limitations of the included studies were use of multiple neoadjuvant treatment protocols within studies, a lack of a description of initial resectability, a lack of a description of the proportion of patients with a pre-treatment biopsy proving malignancy as well as insufficient details regarding intra-operative handling of the vessels. Another major limitation of identified studies is a combination of patient populations with different diagnoses that have different therapeutic options and different outcomes (intrahepatic, hilar, extrahepatic cholangiocarcinoma and even inclusion of other types of biliary tract cancers in the same analysis).

As a result of this systematic review, the authors conclude that there is currently no evidence to suggest improved outcomes with neoadjuvant therapy compared with upfront resection in resectable patients. There is evidence to suggest improved R0 resectability after neoadjuvant therapy and improved survival when neoadjuvant therapy is followed by a surgical resection in primarily unresectable patients. Both of these suggestions come from level 4 evidence.

Given the relatively low incidence of these tumours, well-designed phase II studies should be carried out to study this question. In order to provide guidance to researchers in terms of neoadjuvant strategies, the authors summarized the neoadjuvant therapies from included studies in Table 2 (studies using PDT are not included in the table as PDT has a limited role in downstaging). From the table it becomes obvious that there is no standard protocol or timing of subsequent surgery that should be further studied. An alternative to the described options^{30–34,36,37} or current therapies with well-described response rates in the palliative setting¹⁶ can be the Mayo protocol of neoadjuvant therapy used prior to liver transplantation.²³

Table 2.

Description of neoadjuvant therapy regimens

Study		Chemotherapy	Radiation Therapy	Chemo/RT Timing	Preoperative reassessment	Surgery Timing
1	McMasters et al., 199736	5-FU cont inf (300 mg/m2/day, Mon–Fri)	45 or 50.4 Gy (1.8 Gy DF) 2 pts 30Gy (3Gy DF) 2 pts 10Gy 192-Ir brachy boost 3 pts IORT 10 Gy	concurrent (no. of cycles not described)	not described	not described
3	Gerhards et al., 200034	NA	AP/PA RT 10.5 Gy (3.5Gy DF)	three consecutive days	NA	within 1 week of the last RT fraction, mean 2.9 (±2.0 days)
6	Nelson et al., 200833	flouropyrimidin-based (cont inf n = 26, bolus n = 8, oral n = 10)	3D conformal RT, median dose 50.4 Gy (10.8–54)/1.8–2 Gy DF 4 pts 192-Ir brachytherapy (median dose 29 Gy)	concurrent (except 1 pt), no details about number of cycles described	not described	not described
7	Katayose et al., 201137	Gemcitabin cont inf 400–600 mg/m2	50.4 Gy (range, 10.8–54) 1.8–2 Gy DF	2 weeks prior surgery (5 weeks of chemo/RT)	CT or MRI Timing not described response by RECIST criteria	2 weeks after chemo/RT if no PVE, 3–4 weeks if PVE, range: 15–48 days (median 27 days)
8	Glazer et al., 201232	Gemcitabin/Cisplatin or FU, no standardized protocol	2 pts, no details described	4–5 months – number of cycles described	not described	delayed by median of 6.8 months if chemo/RT
A1	Sano et al., 201131	Gemcitabine cont inf 1000 mg/m2	NA	2 courses	CT	not described
A2	Tada et al., 201230	Gemcitabine and S-1	NA	4 months	CT	not described

DF, daily fraction; pts, patients; cont inf, continuous IV infusion; RT, radiation therapy; CT, computed tomography; NA, not available; PVE, portal vein embolization; RT, radiotherapy.

Future studies should meticulously describe initial diagnosing process, the reason for unresectability and selection criteria for neoadjuvant therapy, the therapy itself including details of surgical procedures such as vascular resections. Outcome descriptions should include rates of R0 resections, complete and partial response rates, post-operative morbidity and survival for each group of patients separately. This would eliminate the majority of limitations of previous studies. Again, given the relatively low incidence it is very likely that systematic reviews and meta-analyses will play a major role in answering more specific questions in the future. Therefore standardization is imperative to enable such research.

Conflicts of interest

None declared.