Introduction

Pancreatic cancer represents one of the leading causes of cancer-related death and the fourth cause of cancer mortality in the USA [1, 2]. The majority of the cases diagnosed with pancreatic cancer are ductal adenocarcinomas (PDAC), commonly located in the head of the pancreas [3, 4]. The disease is generally advanced at the time of diagnosis, is associated with poor prognosis, and a significant number of patients present with a non-resectable tumor [5]. In fact, the 5-year survival rate of pancreatic cancer is approximately 6% [5]. Depending on the degree of differentiation and the tumor microenvironment, the malignancy may present poorly to well-formed glands or infiltrating cells forming sheets [3, 4]. Despite the significant progress in research, the mortality rate regarding pancreatic cancer continues to increase and it is projected that by 2030 pancreatic cancer will be the second cancer-related cause of mortality [6].

Given the poor prognosis of patients with locally advanced pancreatic cancer, it is crucial to develop novel treatment modalities that enhance survival and quality of life. High-intensity focused ultrasound (HIFU) has emerged as a promising non-invasive imaging-guided thermal ablation technique inducing thermal and mechanical energy to the targeted tumor tissue, without affecting the surrounding healthy tissue [7, 8]. HIFU is guided by other imaging modalities such as CT (Computed Tomography guided Focused Ultrasound—CTgFUS) or MRI (Magnetic Resonance guided Focused Ultrasound—MRgFUS) to enable guidance of the treatment and monitoring. Although HIFU delivers thermal ablation, the temperature reached is not high enough to cause immediate cell necrosis [7]. In fact, HIFU induces the intracellular denaturation of proteins and stored pancreatic enzymes, followed by cellular degeneration and necrosis [7]. Except for thermal effects, HIFU also induces mechanical effects associated with high-intensity acoustic energy [7]. Through these effects HIFU in conjunction to systemic chemotherapy is expected to provide pain relief, quality of life enhancement, along with increased survival in patients with locally advanced pancreatic cancer. According to a previous meta-analysis [9], HIFU appears to be an effective tool for pain palliation in this patient group. However, it failed to analyze survival endpoints and its outcomes were associated with high heterogeneity [9]. As the number of studies assessing the feasibility of HIFU for patients with advanced pancreatic cancer increases, it is necessary to conduct a meta-analysis to provide the best level of evidence on the topic. The purpose of this study is to summarize the existing evidence assessing the survival and pain-relief outcomes of HIFU in conjunction with systemic chemotherapy.

Materials and methods

Search strategy and articles selection

The current study was designed and performed according to the protocol agreed by all participating authors and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [10]. A thorough literature search was performed in three databases—Scopus (ELSEVIER), Pubmed (Medline), and Cochrane Central Register of Controlled Studies (CENTRAL) (last search: June 30, 2021) using the following keywords: “high intensity focused ultrasound” or “hifu”; “pancreatic cancer” or “pancreatic adenocarcinoma” or “pdac”; “advanced” or “locally advanced”; “treatment” or “palliation” or “palliative” or “pain” or “survival”. The inclusion criteria of the present meta-analysis were as follows: (1) original articles with > 5 patients, (2) written in the English language, (3) published from 1990 to 2021, (4) conducted on human subjects (5) reporting outcomes (median overall survival (OS), OS at 6 and 12 months, pain relief, CA19-9 reduction, and complications) evaluating the implementation of HIFU in the treatment of patients with locally advanced pancreatic cancer, and (6) in cases of multiple studies reporting on the same population, only the largest study or the one with the longest follow-up was included.

Two investigators (MPF, DEM) independently extracted data from the included studies. Any discrepancies between the two authors regarding the inclusion or exclusion of studies were discussed with the senior author (DZ) so as to include only the articles that best matched the protocol until a consensus was reached. Besides, the reference lists of the included articles were further evaluated for additional potentially eligible articles.

Data extraction

For each study that was included, data were extracted relative to demographics (sample size regarding each group, age, sex, site/size/stage of the tumor, type of image guidance, treatment modality, type of HIFU device), along with the primary endpoints (median overall survival (OS), OS at 6 and 12 months) and secondary endpoints (pain relief, CA19-9 reduction, and complications). Two authors (MPF, DEM) performed the data extraction and compared the validity of the data until a consensus was reached. Additionally, the kappa coefficient test was applied in order to assess the level of agreement between the two investigators.

Statistical analysis

Regarding the categorical outcomes, the Odds Ratio (ORs) and 95% confidence interval (95% CI) were estimated, by means of the Random-Effects model (Mantel–Haenszel statistical method). OR < 1 denoted an outcome that was more frequent in either the post-HIFU treatment evaluation, when compared with the baseline status, or in the control group, when compared with the HIFU group. Continuous outcomes were evaluated by means of weighted mean difference (WMD) with its 95% CI, using Random-Effects (Inverse Variance statistical method) models. In cases where WMD < 0, values in either the post-HIFU treatment evaluation or the control, respectively, were increased. The Random-Effects model was chosen, given that it was not expected that all included studies would share a common effect size. Inter-study heterogeneity was assessed through Cochran Q statistic and by estimating I2 [11]. Forest plots were produced regarding the variables that were analyzed. Data analysis was performed using the Cochrane Collaboration RevMan version 5.4.1.

Quality and publication bias evaluation

The Newcastle–Ottawa Quality Assessment Scale (NOS) [12] was facilitated to evaluate all non-Randomized Controlled Trials (non-RCTs). The scale ranges from zero to nine stars. Studies that were graded with a score equal to or higher than five stars were considered to have adequate methodological quality and were incorporated. The RCTs were assessed for their methodological quality with the tools used to evaluate the risk of bias according to the Cochrane Handbook for Systematic Reviews of Interventions [11]. Two authors (MPF, DEM) rated the included studies independently and the final decision was reached by consensus.

The risk of publication bias was evaluated by the visual inspection of the funnel plots. Publication bias could not be further evaluated using Egger’s formal statistical test [13] due to the inadequate number of the study arms that were included in the analyses, thus substantially compromising the power of the test.

Results

Article selection and patient demographics

The flow diagram of the systematic literature search is shown in Fig. 1. Among the 257 articles that were originally retrieved, nineteen articles were included in the qualitative and eleven in the quantitative synthesis. The level of agreement between the two reviewers regarding data extraction was “almost perfect” (kappa = 0.920; 95% CI 0.831, 1.000). The study design was retrospective in eight studies [15, 18, 20, 21, 24, 27, 28, 32], prospective in ten studies [14, 16, 17, 19, 23, 25, 26, 29,30,31], randomized controlled in one study [22], and were published between 2005 and 2021. The total study population was 939 patients. The image guidance was performed using CT/MRI modalities in eight studies [14,15,16,17,18, 24, 27, 28] and US in eleven studies [19,20,21,22,23, 25, 26, 29,30,31,32]. Baseline characteristics and information regarding tumor characteristics, along with the type of treatment, and HIFU characteristics are presented in Table 1. The pooled estimates of primary and secondary endpoints are reported in Table 2, and the reported complications in Table 3. The Newcastle–Ottawa Scale (NOS) assessment for all studies is shown in Table 1.

Fig. 1
figure 1

Trial-flow of the systematic review and meta-analysis

Full size image
Table 1 Baseline characteristics of the studies that were finally included in the meta-analysis
Full size table
Table 2 Summary of the pooled estimates regarding the outcomes of the included patients
Full size table
Table 3 Summary of the main complications of HIFU treatment
Full size table

Primary endpoints: survival

All studies included in the analyses of the primary endpoints compared patients treated with either HIFU plus chemotherapy (gemcitabine) or chemotherapy (gemcitabine regimen) alone. The median OS was also higher in the HIFU group compared with the chemotherapy-alone group (WMD: 2.83 [95% CI 1.06, 4.59]; p = 0.002 (Fig. 2a, Table 2). The OS at 6, and 12 months was higher (p < 0.05) in patients treated with HIFU combined with chemotherapy compared with those receiving standard chemotherapy alone (Fig. 2b, c, Table 2). Although it would be of great interest to compare HIFU plus chemotherapy with HIFU alone treatment, no relative data were available to perform such a comparison.

Fig. 2
figure 2

Forest plot describing the differences in a median overall survival (OS), b OS at 6 months, and c OS at 12 months

Full size image

Secondary endpoints: pain relief and tumor responsiveness

The reported pain was assessed using the numerical rating score (NRS). The reported level of pain was significantly lower after HIFU treatment (WMD: 2.54 [95% CI 1.91, 3.16]; p < 0.001) (Fig. 3a, Table 2). CA19-9 was significantly reduced following HIFU treatment (WMD: 49.81 [95% CI 8.53, 91.09]; p = 0.02) (Fig. 3b, Table 2).

Fig. 3
figure 3

Forest plot describing the effect of HIFU on a pain relief and b CA19-9 levels

Full size image

Secondary endpoints: complications

Complications are presented in Table 3. Complications were reported in all the included studies. The most common reported complication was skin burn. Other minor complications included abdominal pain, vomiting, fever, and local edema. These side-effects were treated conservatively and were resolved in a period of 6 weeks. Major adverse events were rare and included portal vein thrombosis (one case [25]), severe abdominal pain requiring hospitalization, pancreatic fistula (two cases [26]), and jaundice aggravation (four cases). No death was reported.

Publication bias

The evaluation of median OS and pain levels were associated with increased heterogeneity (Table 2). Furthermore, the heterogeneity of tumor responsiveness in terms of Ca19-9 levels, along with OS at 6 and 12 months post-intervention was generally low. Funnel plots (Fig. S1) seemed asymmetrical, with studies being absent from either top or bottom of the graph, thus suggesting certain publication bias. The small number of the included studies was the main reason for the reported asymmetry.

Discussion

Despite the progress in cancer research, survival and quality of life of patients with pancreatic cancer remain poor [5]. In fact, locally advanced pancreatic cancer is associated with reduced median OS, while the quality of life is affected by pain [5]. In fact, the pain with pancreatic cancer origin is multifactorial, and is mainly attributed to tumor infiltration of nerves, along with compression and inflammatory reaction elicited by the malignancy [33]. In this context, the exact physiologic mechanisms by which HIFU reduced the reported pain are not clear, yet. There have been proposed three potential mechanisms of action: (1) thermal collateral damage to the nerves adjacent to the tumor, (2) the reduced post-ablation size of the tumor resulting in reduced mass effect, and (3) the ablation of celiac plexus fibers [34]. Compared with a previous meta-analysis [9], the present study is the first to analyze survival endpoints, thus providing the best level of evidence on the topic.

According to our meta-analysis, HIFU is associated with an increased survival when combined with chemotherapy. In fact, our outcomes indicate enhanced mean OS, along with survival at 6 and 12 months post-HIFU treatment in patients receiving combined therapy compared with those receiving only systemic chemotherapy. The physiologic background that explains these findings is that PDAC is a relatively hypovascular tumor, enclosed by a fibrous ring that limits the penetration and diffusion of chemotherapeutic agents. In this context, HIFU might provide a synergistic effect with chemotherapy, through increased vascular permeability, thus boosting the drug concentration in the tumor microenvironment [8]. Furthermore, our study indicated that HIFU inhibits tumor progression, as evaluated by CA19-9 levels and tumor volume. In fact, the present meta-analysis represents the first evidence synthesis indicating the value of HIFU combined with chemotherapy in raising OS of patients with advanced pancreatic cancer.

Originally, HIFU was presented as a palliative treatment for pain relief in patients with advanced pancreatic cancer. In the present meta-analysis, pain relief was evaluated as a secondary endpoint. According to our outcomes, HIFU is a very effective mean of relieving pain in patients with advanced pancreatic cancer. Differences regarding the treatment protocol among different centers contributed to high heterogeneity. Nonetheless, to limit the potential publication bias we excluded studies with a small number of patients, along with studies incorporating the same population. Finally, the evidence provided by case reports that were excluded from the present meta-analysis are consistent with our findings in terms of efficacy and safety of HIFU for pain palliation [35, 36]. The duration of pain relief has been validated for distant follow-up periods up to 17 months [30].

Complications represent a major endpoint when evaluating a potential treatment intervention in patients with advanced cancer, given that it is crucial to preserve the adequate quality of life standards. Herein, we demonstrated that HIFU is not associated with significant morbidity. In fact, the main alternate proposed treatment when opioids fail is neurolytic celiac plexus blockade (NCPB) [33]. However, NCPB is an invasive procedure, with only short-term efficacy (up to 3 months) and is associated with significant adverse events, such as local pain, diarrhea, hypotension, pneumothorax, and neurological side-effects [37].

The majority of the studies included in the present systematic review used US-guided HIFU technique, while eight studies implemented a CT/MRI-guided HIFU approach. The US-guided approach implements the ultrasound for both detecting and ablation of the lesions, allowing identification of potential obstructions in the US beam pathway. Nonetheless, it is associated with certain limitations regarding the quality of depiction of tumor and its borders, the lack of temperature monitoring, along with its operator-dependent attributes. In this context, the contrast-enhanced US could provide enhanced capabilities of tumor imaging, thus raising the feasibility and efficacy of HIFU. MRI-guided HIFU has been proposed as an enhanced approach in terms of efficacy. However, there is a lack of real-world evidence on its superiority, cost-effectiveness, and availability compared with US-guided method.

The limitations of the present meta-analysis reflect the limitations posed by the included studies. The majority of the studies were prospective, eight studies were retrospective, and only was RCT. The small number of arms in the meta-analyses poses also a certain publication bias, as it reflects the asymmetry of the funnel plots. The differences between the two groups and the high heterogeneity among studies regarding the baseline characteristics, the primary tumor location and biology, extend of the disease, the therapeutic protocols, along with the HIFU specifications should also be taken into consideration. Moreover, most of the included studies did not clearly state the dropout rate of patients during the protocol pathway, which is crucial in terms of evaluation of the success regarding each strategy. Furthermore, there was a lack of homogeneity among studies regarding the classification of complications, since classification scales such as Clavien–Dindo were not adopted. In addition, the available data were limited regarding the primary endpoints (median OS, OS at 6 and 12 months), thus posing a certain limitation in the present meta-analysis. Nonetheless, the limited available evidence highlights the value of the present meta-analysis as the best currently available level of evidence. Therefore, there is a need for well-designed RCTs with standardized reporting of staging, therapeutic protocols, HIFU specifications, and outcomes including dropout rate, pain relief, complications according to a classification scale, median and long-term OS and DFS.

On the contrary, the strengths of the present meta-analysis are as follows: (1) the clear study protocol, (2) the well-described inclusion criteria, (3) the systematic literature search in three different databases, (4) the quality assessment of the included studies (5) the detailed presentation of the outcomes, and (6) the low heterogeneity regarding the primary and secondary endpoints.

Conclusion

The present meta‐analysis suggests that HIFU represents a safe and feasible fourth treatment modality for patients with advanced pancreatic cancer, in conjunction with surgery, chemotherapy, and radiotherapy. In this context, the aim of oncologic care should be to evaluate each patient individually by the multi-disciplinary team and to select the appropriate strategy for each case. Patients with a higher burden of pancreatic disease, presenting with severe pain, might possibly be candidates for a HIFU approach, whereas the classical strategy may be more suitable for patients with a stable disease and with low levels of pain. Nonetheless, new RCTs with reduced selection bias and increased clarity of significant outcomes are required.