The clinical benefit of hyperthermia in pancreatic cancer: a systematic review

Abstract

Objective: In pancreatic cancer, which is therapy resistant due to its hypoxic microenvironment, hyperthermia may enhance the effect of radio(chemo)therapy. The aim of this systematic review is to investigate the validity of the hypothesis that hyperthermia added to radiotherapy and/or chemotherapy improves treatment outcome for pancreatic cancer patients.

Methods and materials: We searched MEDLINE and Embase, supplemented by handsearching, for clinical studies involving hyperthermia in pancreatic cancer patients. The quality of studies was evaluated using the Oxford Centre for Evidence-Based Medicine levels of evidence. Primary outcome was treatment efficacy; we calculated overall response rate and the weighted estimate of the population median overall survival (m_p) and compared these between hyperthermia and control cohorts.

Results: Overall, 14 studies were included, with 395 patients with locally advanced and/or metastatic pancreatic cancer of whom 248 received hyperthermia. Patients were treated with regional (n = 189), intraoperative (n = 39) or whole-body hyperthermia (n = 20), combined with chemotherapy, radiotherapy or both. Quality of the studies was low, with level of evidence 3 (five studies) and 4. The six studies including a control group showed a longer m_p in the hyperthermia groups than in the control groups (11.7 vs. 5.6 months). Overall response rate, reported in three studies with a control group, was also better for the hyperthermia groups (43.9% vs. 35.3%).

Conclusions: Hyperthermia, when added to chemotherapy and/or radiotherapy, may positively affect treatment outcome for patients with pancreatic cancer. However, the quality of the reviewed studies was limited and future randomised controlled trials are needed to establish efficacy.

Keywords:

• HT treatment
• hyperthermic
• radiative heating
• thermal dosimetry
• thermal dose

Introduction

The prognosis for pancreatic cancer is very poor and has scarcely improved over the last two decades, with a 5-year overall survival of less than 10% [1,2]. For (borderline) resectable pancreatic cancer, the standard treatment is exploratory laparotomy, with resection and adjuvant chemotherapy when a resection was performed [3–5]. The impact of neoadjuvant chemoradiotherapy is under investigation; its suggested beneficial effect on overall survival has been studied in two randomised trials that were terminated prematurely [6,7], as well as in one recently completed randomised multicenter trial [8]. However, the vast majority of patients present with either locally advanced or metastatic pancreatic cancer [9]. Also, most patients develop recurrent disease after surgery. For both groups, chemotherapy is the treatment of choice [10].

While chemotherapy and/or radiotherapy are successful in many tumour types, pancreatic cancer remains quite resistant to these conventional therapies. This resistance is in large part due to the abundant and dense accumulation of nontumour cells known as the stroma, which obstructs vascularisation and the delivery of chemotherapeutics. In addition, it hampers the delivery of oxygen, causing hypoxia [11]. Such a hypoxic microenvironment diminishes sensitivity to chemotherapeutics and radiation and negatively impacts prognosis [12–16].

To improve outcome, new approaches are being explored. For clinically fit patients with locally advanced (LAPC) or metastatic pancreatic cancer, the multiagent regimen FOLFIRINOX (i.e. leucovorin, fluorouracil, irinotecan and oxaliplatin) has shown promising results [17,18]. The combination of gemcitabine and nab-paclitaxel showed improved survival over gemcitabine monotherapy in patients with metastatic pancreatic cancer [19,20]. For localised disease, encouraging results from stereotactic body radiation therapy (SBRT), radiofrequency ablation (RFA) and irreversible electroporation (IRE) have been reported, though evidence from randomised trials is still lacking [21–24]. However, the aforementioned treatments are often taxing to the patient, have high toxicity rates, or are still lacking evidence of efficacy. Therefore, development of treatment options that may directly reduce resistance to the conventional therapies is an attractive approach, for example, by counteracting the negative impact of the tumour stroma.

Mild hyperthermia (HT), that is, raising the temperature of the tumour to approximately 40–44 °C, typically for one hour, acts as a sensitiser for radiotherapy and chemotherapy [25–27]. In this temperature range, improved drug delivery and dispersion in the tumour enhances the effect of chemotherapy [28–30]. With improved blood flow, hypoxia is decreased, which further enhances radiation effects through an increased production of oxygen radicals. In addition, hyperthermia inhibits DNA repair, thus enhancing tumour cell killing [31,32].

Efficacy of HT, in combination with radiotherapy, chemotherapy or chemoradiotherapy, has been demonstrated for several tumour types [26,33,34]. For pancreatic cancer, the efficacy of HT has been validated in the preclinical setting. For example, in subcutaneous pancreatic cancer xenografts in rats and mice, the combination of gemcitabine and HT was more effective than gemcitabine or HT alone [35–37]. For pancreatic cancer patients, however, data are limited.

The objective of this systematic review is to investigate the literature on the effect of hyperthermia combined with radiotherapy and/or chemotherapy for pancreatic cancer. We aim to test the validity of the hypothesis that the addition of hyperthermia as a sensitiser to radiotherapy and/or chemotherapy for locally advanced and metastatic pancreatic cancer improves overall survival.

Methods

Execution of this systematic review was based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standard guidelines [38]. The review was registered at PROSPERO, the international prospective register of systematic reviews (registration number CRD42017073851).

Search strategy

MEDLINE and Embase were systematically searched from onset using the Ovid interface. Key concepts were “hyperthermia” and “pancreatic cancer”. For each concept, eligible synonyms and controlled terms were identified by a scoping search. The systematic search was supplemented with handsearching (Figure 1). See Appendix A for a detailed display of the systematic Ovid search strategies.

Figure 1. PRISMA diagram. OS: overall survival; other language = Chinese, Japanese or Russian; ^a[68] describes same study as included study [57]. ^bcohort in [67] largely overlaps with cohort in included study [54].

Selection

Selection of articles was independently performed by two reviewers (EV and AVDH) using Rayyan, a web application to screen identified articles [39]. In case of discrepancies, consensus was reached in deliberation with a third person (JC). The first selection was based on title and abstract; the second selection was based on the full text of the articles.

Inclusion criteria

Papers were included according to the following criteria: pancreatic cancer; study conducted in humans; number of patients >1; disease stage described; HT treatment aimed at a temperature in the range 40–45 °C in combination with radiotherapy and/or chemotherapy; treatment regimen described (type and dose of chemotherapy; radiation dose; type, temperature and duration of HT); outcome of HT treatment described in terms of overall survival (OS) and/or tumour response; text in English, German, French or Dutch language.

Exclusion criteria

Studies involving hyperthermic intraperitoneal chemotherapy (HIPEC) and thermal ablation were not included. Studies that included patients with a variety of carcinomas were only included when they reported on the subgroup of pancreatic cancer patients separately. Studies without a control cohort were excluded when the treatment regimen was different than commonly prescribed (e.g. combinations of compounds not in current clinical use). Since studies applying hyperthermia in pancreatic cancer were likely to be scarce, no particular level of evidence was required for inclusion.

Data extraction and analysis

For the selected studies, we assessed type of study, patient cohorts, delivered treatments, treatment efficacy and reported toxicity. Therefore, the following data were extracted from the articles (for the study group and, when applicable, also for the control group): study characteristics; number of patients; median age; tumour stage; tumour type and histological confirmation; previous treatments; treatment protocol (HT treatment regimen, complementing chemotherapy and/or radiotherapy treatment regimen); type of HT; type of temperature measurement; treatment protocol of control group, when available; tumour response; overall survival; reported side effects, grade ≥3. When the median overall survival was not stated, but a Kaplan–Meier survival plot was supplied, the median OS was obtained from that plot by determining the time on the x-axis at 50% survival rate.

Evaluation of study quality and risk of bias assessment

The quality of the included studies regarding treatment benefits of HT was evaluated according to the 2011 Oxford Centre for Evidence-Based Medicine (OCEBM) levels of evidence [40].

In addition, we assessed risk of bias based on the following criteria: the presence of control group, randomness of allocation, comparability of prognostic parameters (age, stage, etc.) and follow-up, selective reporting and confounding by indication.

Patients characteristics and treatment regimens

To obtain insight into the patient cohorts included in the studies, we tabulated median patient age as well as tumour stage. For the latter, we distinguished between locally advanced (2012 AJCC stage III [41]) and metastatic disease (stage IV). The definition of the AJCC staging has changed over the years. Therefore, when necessary and possible, we reported staging according to the current AJCC guidelines.

Most pancreatic cancers are pancreatic ductal adenocarcinomas (PDACs) [42]. However, confirmation by means of histology is not always obtained. For each study, we assessed the number of PDAC patients and whether histological and/or cytological confirmation was obtained. As treatments may differ between studies, we presented treatment as well as previously received treatments.

Temperature measurements

In HT, the thermal dose and the resulting temperature obtained in the tumour are crucial. While radiosensitisation by increased perfusion and reoxygenation [43,44] can be observed already at 39 °C, the effect increases substantially with increasing temperature with more direct radiosensitisation due to the inhibition of DNA damage repair machinery [45–47]. Also chemosensitisation starts at relatively low temperatures for many chemotherapy compounds, including drugs frequently used in pancreatic cancer, such as gemcitabine and fluorouracil [48–51]. For other drugs, the effect appears to be additive and only effective at higher temperatures [32].

Tissue heterogeneity, perfusion and the presence of (large) blood vessels can greatly affect local temperatures. Therefore, it is paramount that the temperature is properly monitored, preferably in the target volume. However, for pancreatic cancer intratumoral measurements are often not performed, due to the risks associated with such invasive measurements.

Here, we examined whether and how temperature was measured and whether these measurements were intratumoral. Furthermore, we determined per study the intended temperature as well as the achieved temperature or temperature range.

HT efficacy

To investigate efficacy of hyperthermia as adjuvant therapy in pancreatic cancer patients, we analysed response rate, reported as complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD). From this, we calculated the overall response rate over the studies as (CR + PR)/n, with n the combined total number of patients in those studies.

In addition, we examined the median OS. Unbiased pooled estimates of median survival times cannot be achieved by simple weighted averaging of medians; therefore, we calculated m_p, a weighted estimate of the population median overall survival, using [52]: (1) with m_i the median OS of study population i, k the number of included studies, and n_i/N the weight function with n_i the number of patients in study i and N the total number of patients in studies 1–k. For the studies that included a control cohort, we calculated m_p for the HT cohorts and the control cohorts. In addition, we calculated m_p for the HT cohorts over all studies.

HT toxicity

We investigated whether toxicity was reported according to established guidelines and whether HT-related toxicity was separately reported. For each study, we registered the reported toxicity ≥ grade 3.

Results

Literature search

A total of 1293 potentially eligible articles were identified. After the first and second selection (Figure 1), 14 studies were included [53–66]. Two patient cohorts were each described in two articles; for these we excluded the oldest [67,68] and included the latest [54,57]. Based on the reported patient characteristics in [56], it was evident there was no overlap of this patient cohort with the cohort in references [57,68]. One included article was written in German [58]; all others were written in English.

The included 14 studies combined 395 pancreatic cancer patients of whom 248 received HT (Table 1). Six out of 14 studies used a control population that enabled a comparison between treatments with and without hyperthermia [54,55,57,60,62,66]. In two studies, patients with a variety of carcinomas were included, but data were reported for the pancreatic patients separately [53,59].

Table 1. Patient characteristics. Number of patients, median age, disease stage per study, fraction of adenocarcinomas and whether histological confirmation was obtained.

		Median age (year)	Locally advanced disease (stage III)		Metastatic disease (stage IV)			Histology
Study	n	Total (HT/control)	HT	Control	HT	Control	PDAC	Confirmation
Kouloulias et al. 2002 [57]	37	62 (62/60)	6/10 (60%)	19/27 (70%)	4/10 (40%)	8/27 (30%)	37/37	Yes
Ohguri et al. 2008 [60]	29	64 (62/65)	20/20 (100%)	9/9 (100%)	0/20 (0%)	0/9 (0%)	29/29	14/29
Yamada et al. 1992 [54]	73	60.6^b (NR/NR)	11/17 (65%)^d	32/56 (57%)^d	6/17 (35%)	24/56 (43%)	43	52/73
Maluta et al. 2011 [62]	68	64.3 (66.6/65.7)	34/40 (85%)	26/28 (93%)	6/40 (15%)^f	2/28 (7%)^f	68/68	Yes
Ashayeri et al. 1993 [55]	24	NR (62/61)	5/5 (100%)	19/19 (100%)^e	0/5 (0%)	0/19 (0%)	24/24	NR
Maebayashi et al. 2017 [66]	13	78 (62/NR)	5/5 (100%)	8/8 (100%)	0/5 (0%)	0/8 (0%)	4/13	4/13
Douwes 2006 [58]	30	59.8 (59.8/–)	5/30 (17%)^g	–	25/30 (83%)	–	NR	NR
Tschoep-Lechner et al. 2013 [64]	23	60 (60/–)	2/23 (9%)	–	21/23 (91%)	–	22/23	Yes
Volovat et al. 2014 [65]	19	NR (NR/–)	0/19 (0%)	–	19/19 (100%)	–	11/19	Yes
Ishikawa et al. 2012 [63]	18	64 (64/–)	6/18 (33%)	–	12/18 (67%)	–	18/18	Yes
Bakshandeh-Bath et al. 2009 [61]	13	57 (57/–)	0/13 (0%)	–	13/13 (100%)	–	13/13	Yes
Kouloulias et al. 2001 [56]	7	65 (65/–)	7/7 (100%)	–	0/7 (0%)	–	7/7	Yes
Kakehi et al. 1990 [53]^a	34	NR (NR/–)	6/34 (18%)	–	28/34 (82%)	–	23/34	23
Bull et al. 2008 [59]^a	7	62 (62/–)	0/7 (0%)	–	7/7 (100%)	–	NR	Yes
Overall	395	57–78^c	107 (43%)	113 (77%)	141 (57%)	34 (23%)

n: number of patients; –: not applicable; NR: not reported.

^a The pancreatic cancer patients are part of a larger cohort [53,59].

^b Average age.

^c Range of median age.

^d This includes 12 patients (4 HT, 8 control) with stage I and II.

^e Control cohort is a comparable group treated a decade earlier than the HT cohort.

^f Note: for the overall survival analysis, the 8 patients with metastatic disease were excluded [62].

^g Four patients had stage II disease.

Study quality and risk of bias

None of the studies were randomised controlled trials; eight were retrospective studies. Six studies had a control group; in one of these studies, the control group was a cohort of patients treated ten years earlier [55]. Thus, the OCEBM level of evidence was level 3 for five studies and level 4 for all others. Risks of bias are described in Supplementary Table S1.

Patient characteristics and treatment regimens

Most of the 395 pancreatic cancer patients had either LAPC (204 patients, 52%) or metastatic disease (175 patients, 44%); in two studies, a total of 16 patients had resectable or borderline resectable pancreatic cancer [54,58], Table 1. Seven studies included both patients with locally advanced and patients with metastatic disease. Not every study reported on previous treatments received by the patients. For the 201 patients who received hyperthermia in the 12 studies that did report on previously received treatment, 10% had received resection, 23% chemotherapy, 1% chemoradiotherapy and 66% had received no previous treatment before enrolling in a study. For the patients in the control cohorts, 21 of the 147 patients (14%) had received resection; all others had received none.

In the studies, a wide variety of treatments were used (Table 2), with differences in HT treatment (method, temperature, duration, number of sessions) and in (concurrent) chemotherapy (drug, dose) and radiotherapy (dose, fractionation). For the patients who received HT within the studies, this HT was in combination with radiotherapy for 17 patients (7%, two studies) [53,55], with chemoradiotherapy for 82 patients (33%, four studies) [54,60,62,66], and with chemotherapy for 149 patients (60%) in the remaining eight studies. Three types of hyperthermia treatment were applied: locoregional, whole-body and intraoperative hyperthermia.

Table 2. Treatment regimens.

	HT				Control
			Treatment within study				Treatment within study
Study	n	Previous treatments	HT treatment	Adjuvant treatment	n	Previous treatments	Treatment	Adjuvant treatment
Locoregional HT
Kakehi et al. [53]	34	NR	c-LRHT (40–50 min.) + 50–60 Gy RT (n = 12); c-LRHT (40–50 min.)+ MMC +5-FU derivative (n = 22)		–	–	–
Douwes [58]	30	Resection (n = 14); none (n = 16)	c-LRHT (60 min.) + 5-FU every 4 weeks		–	–	–	–
Ohguri et al. [60]	20	Resection (n = 4)	c-LRHT (30–50 min.) + gem + RT (gem prior to (n = 6); gem concurrent with (n = 14))		9	Resection (n = 2)	Gem
Maluta et al. [62]	40	None	r-LRHT (60 min.) + 30–66 Gy RT + gem (+ CDDP (n = 7); + 5-FU (n = 3); + oxali (n = 1))		28	None	30–66 Gy RT + gem (+ CDDP (n = 5); + 5-FU (n = 3); + oxali (n = 1))
Ishikawa et al. [63]	18	None	c-LRHT (40 min.) + gem		–	–	–	–
Tschoep-Lechner et al. [64]	23	Gem (n = 22); 5-FU + RT (n = 1)	r-LRHT (1 h) + gem + CP	5-FU (n = 1); gem + Tarceva (n = 2); xeloda (n = 2); ixoten (n = 1); RT + gem + CP (n = 1)	–	–	–	–
Volovat et al. [65]	19	Gem	c-LRHT (60 min.) + gem-oxali		–	–	–	–
Maebayashi et al. [66]	5	None	c-LRHT (50 min.) + gem +50–60 Gy RT		8	None	Gem +50–60.8 Gy RT (n = 3); 5-FU +50–60.8 Gy RT (n = 5)
Intraoperative HT
Yamada et al. [54]	17	Curative resection^a (n = 2)	c-IOHT (1 h) + 20–35 Gy IORT	postop 30–45 Gy RT (n = 15), 3 or 4 sessions postop c-LRHT (n = 11); most patients received chemo (MMC, 5-FU, other)	56	Curative resection^a (n = 19)	20–25 Gy IORT	Postop 30–45 Gy RT (n = 5); most patients received chemo (MMC, 5-FU, other)
Ashayeri et al. [55]	5	None	r-IOHT (40 min.) + 20–25 Gy IORT	postop 40 Gy RT	19	None	20–25 Gy IORT	Postop 40 Gy RT
Kouloulias et al. [56]	7	None	r-IOHT + IO 5-FU	preop 5-FU; postop 5-FU +45 Gy RT	–	–	–	–
Kouloulias et al. [57]	10	None	r-IOHT (1 h) + IO 5-FU	preop 5-FU, postop chemo (5-FU + DOX + CP) + 45 Gy RT	27	None	IO 5-FU	Preop 5-FU, postop chemo (5-FU + DOX + CP) + 45 Gy RT
Whole-body HT
Bull et al. [59]	7	Gem (n = 2); CP/gem (n = 2); gem + carbo (n = 1); gem + RT + gene therapy (n = 1); none (n = 1)	rh-WBHT (6 h) + CP + gem + IFN-α		–	–	–	–
Bakshandeh-Bath et al. [61]	13	Most: gem	rh-WBHT (60 min.) + gem + carbo		–	–	–	–
Overall	248				147

c: capacitive; CP: cisplatin; carbo: carboplatin; oxali: oxaliplatin; DOX: doxorubicine; 5-FU: fluorouracil; gem: gemcitabine; HT: hyperthermia; IFN-α: interferon alpha; IOHT: intraoperative HT; IORT: intraoperative RT; LRHT: locoregional HT; n: number of patients; –: not applicable; r: radiative; rh: radiant heat; RT: radiotherapy; WBHT: whole body HT.

^a the patients who had resection as previous treatment were not included in the response analysis.

Locoregional hyperthermia was used in eight studies, with heating of tumour and surrounding tissue either radiative (using radiofrequency waves; 63 patients) [62,64] or capacitive (using electrodes; 126 patients) [53,58,60,63,65,66]. Treatment times varied from 40–60 min.

Intraoperative hyperthermia (four studies, 39 patients) was performed using a radiofrequency device to locally heat the tumour and surrounding area for 40–60 min. during surgery [54–57].

Finally, whole-body hyperthermia (two studies) involved heating the entire body to a temperature of 40 °C for 6 h (seven patients) [59] or 41.8 °C for 1 h (13 patients) [61] using a radiant heat device.

Temperature measurements

Temperature was measured during HT treatment in eight of the 14 studies (Table 3). Of these, only three studies measured temperature intratumorally [54,56,57], all using (micro)thermocouples. In the other five studies, gauges were placed (relatively) nearby in, for example, duodenum, bladder or rectum. In the studies in which no temperature was measured, the authors relied on previously established optimal settings for their equipment.

Table 3. Temperature measurements during HT treatment. For each study, it is indicated whether and for how many patients the temperature was measured, how this was done, what the target temperature was and what temperatures were obtained.

Study	T measured	n	Intratumoural	Thermometry	Target T (°C)	Measured T (°C)
Kakehi et al. [53]	Yes	34/34	No	Teflon-coated μTCs, in duodenum or gastric antrum	>42	Tmean 40.5; Tmax 43.5
Douwes [58]	No	–	–	–	42–44	–
Ohguri et al. [60]	No	–	–	–	≥41	–
Maluta et al. [62]	Yes	40/40	No	Thermometer in rectal lumen	≥42	Mean T90 = 40.5 (CI: 39.8–41); mean Tmax =41.1 (CI: 40.2–42.5)
Ishikawa et al. [63]	No	–	–	–	NR	–
Tschoep-Lechner et al. [64]	Yes	11/23	No	Interluminal catheters in bladder, rectum and vagina; measurements every cm along catheter every 5 min^a	42	Mean: 38.5
Volovat et al. [65]	No	–	–	–	41–42.5	–
Maebayashi et al. [66]	No	–	–	–	>41	–
Yamada et al. [54]	Yes	17/17	Yes	3 or 4 Teflon-coated μTCs in tumour and subcutaneous tissue	42.5	>42 in 5/17 patients
Ashayeri et al. [55]	NR	–	–	–	42.5–43	–
Koukoulias et al. [56]	Yes	7/7	Yes	TC probe with 3 measuring points inside 15-gauge flexiguide catheter	43–45	Tmin: 42.4–43.8
Kouloulias et al. [57]	Yes	10/10	Yes	TC probe with 3 measuring points 0.8 cm apart inside 15-gauge flexiguide catheter	43–45	Tmin: median/mean 42.9 (42.4–43.7) Tmax: 44.8 (44.1–45.9)
Bull et al. [59]	Yes	7/7	No	Bladder, rectal, axillary and skin temperatures measured with thermistor probes	40	40
Bakshandeh-Bath et al. [61]	Yes	13/13	No	Rectal and/or oesophageal probe (skin and ambient T also measured)	41.8	41.8 °C ± 0.2 °C (Tmax for rectal or oesophageal probe)
Overall	8/14		3/8

CI: 95% confidence interval; –: not applicable; NR: not reported; TC: thermocouple; μTC: microthermocouple; T: temperature; T_max: maximum T; T_min: minimum T; T90: T in 90% of the measuring locations.

^a Data were compared to data from an earlier group of abdominal and pancreatic cancer patients in same institute for whom intratumoral temperatures were measured.

Efficacy: response rate and overall survival

Response rate was reported in 11 studies (243 patients; Table 4). Complete response was observed for six patients in three studies [53,58,60]; all six had received HT. For the 11 studies, the combined overall response rate for the 192 patients treated with HT was 31.3% (6 CR, 54 PR). For the three studies that included a control group [54,57,60], the response rate for 41 patients treated with hyperthermia was 43.9% (1 CR, 17 PR), compared with 35.3% (18 PR) in the 51 control patients.

Table 4. Tumour response.

		HT				Control
Study	n	CR(%)	PR(%)	SD(%)	PD(%)	CR(%)	PR(%)	SD(%)	PD(%)
Kouloulias et al. [57]	37	0	6 (60%)	4 (40%)	0	0	10 (37%)	14 (52%)	3 (11%)
Ohguri et al. [60]	29	1 (5%)	4 (20%)	15 (75%)	0	0	2 (22%)	6 (67%)	1 (11%)
Yamada et al. [54]	26^b	0	7 (64%)	4 (36%)	0	0	6 (40%)	7 (47%)	2 (13%)
Maluta et al. [62]	68	NR	NR	NR	NR	NR	NR	NR	NR
Ashayeri et al. [55]	24	NR	NR	NR	NR	NR	NR	NR	NR
Maebayashi et al. [66]	13	NR	NR	NR	NR	NR	NR	NR	NR
Douwes [58]	30	1 (3%)	10 (33%)	12 (40%)	7 (23%)	–	–	–	–
Tschoep-Lechner et al. [64]	16^c	0	1 (6%)	7 (44%)	8 (50%)	–	–	–	–
Volovat et al. [65]	19	0	4 (21%)	9 (47%)	6 (32%)	–	–	–	–
Ishikawa et al. [63]	18	0	2 (11%)	9 (50%)	7 (39%)	–	–	–	–
Bakshandeh-Bath et al. [61]	13	0	3 (23%)	5 (38%)	5 (38%)	–	–	–	–
Kouloulias et al. [56]	7	0	4 (57%)	3 (43%)	0	–	–	–	–
Kakehi et al. [53]^a	34	4 (12%)	8 (24%)	18 (53%)	4 (12%)	–	–	–	–
Bull et al. [59]	7	0	5 (71%)	1 (14%)	1 (14%)	–	–	–	–
Overall	236^d	6 (3%)	54 (29%)	87 (47%)	38 (21%)	0	18 (35%)	27 (53%)	6 (12%)

CR: complete response; –: not applicable; NR: not reported; PD: progressive disease; PR: partial response; SD: stable disease.

^a [53]: 22/34 received chemotherapy and 12/34 received RT; response was supplied per treatment cohort but here reported combined.

^b Tumour response was analysed only in 26 out of 73 cases; patients who had undergone tumour resection (n = 27) and for whose there was insufficient data to compare before and after treatment (n = 20) were excluded [54].

^c Only patients with available CT scans were included [64].

^d The three studies that did not report response are excluded from the total number of patients.

Overall survival rates were available for 12 studies (Table 5); two of these studies did not describe median OS for each cohort but did supply Kaplan–Meier curves [54,66]. For the six studies with a control group, the weighted estimate of the population median overall survival m_p was 11.7 months for the cohorts receiving HT (median OS range 6–18.6 months), compared with 5.6 months for the control cohorts (4–11 months). All six studies with control patients showed a longer median OS with HT treatment than without. For the 12 cohorts receiving HT for which median OS was available, m_p was 10.5 months (Table 5).

Table 5. Median overall survival, for patients who received hyperthermia (HT cohort) and patients who did not (control cohort).

	HT			Control
	Overall survival (mo)			Overall survival (mo)
Study	n	Median	Range	n	Median	Range
Kouloulias et al. [57]^a	10	11	5–33	27	7	4–24
Ohguri et al. [60]^a	20	18.6	4–46	9	9.6	5–20.5
Yamada et al. [54]^a,^b	17	6	1–36	56	4	0–15
Maluta et al. [62]^c	34	15	6–20	26	11	5–13
Ashayeri et al. [55]	5	15	5–31	19	5	2–15
Maebayashi et al. [66]^a,^b	5	15	12–18	8	9	5–16
Overall (6 studies)	91	11.7^d	1–46	145	5.6^d	0–24
Douwes [58]	30	8	2–53	–	–	–
Tschoep-Lechner et al. [64]^a	23	12.9	3–35	–	–	–
Volovat et al. [65]^a	19	8.9	5–14.5	–	–	–
Ishikawa et al. [63]^a	18	8	1.5–22.5	–	–	–
Bakshandeh-Bath et al. [61]^a	13	11.4	2.5–39	–	–	–
Koukoulias et al. [56]	7	18.5	7–35	–	–	–
Kakehi et al. [53]	–	NR	NR	–	–	–
Bull et al. [59]	–	NR	NR	–	–	–
Overall (14 studies)	201	10.5^d	1–53

–: not applicable; n: number of patients in cohort; NR: not reported.

^a (Part of) OS range(s) extracted from Kaplan–Meier survival plot.

^b Median OS value(s) extracted from Kaplan–Meier survival plot.

^c The 8 patients with metastatic disease were excluded from OS analysis, thus n = 60 (34 HT; 26 control) [62].

^d Weighted estimate of the population median overall survival m_p (Equation (1)(1) [52].

Reported toxicity

Twelve studies reported on toxicity (Table 6), of which 10 specified the toxicity criteria used (WHO criteria (2 studies) and a version of the NCI-CTCAE (8 studies)). Most studies did not make a distinction between HT-related toxicities and other toxicities. Grade 5 toxicity was not reported in any study; most reported grade 3/4 toxicities were hematologic, likely due to chemotherapy. The only severe hyperthermia-related toxicity was severe subcutaneous fatty burn following intraoperative hyperthermia in one patient [54].

Table 6. Reported toxicity. For each study it is indicated which if any reporting guidelines were used and which graded (grade ≥3) or other toxicities were reported.

Study	Reporting guidelines	Graded toxicity, grade ≥3 only	Otherwise reported toxicity
Kouloulias et al. [57]	NR	None	ChT-related: 7 myelosuppression, 2 increase in BUN and/or creatinine, 4 increases in SGOT and/or SGPT, 9 vomiting, 8 asthenia. All patients: mild gastrointestinal symptoms and alopecia. (= for HT cohort of 10 [23])
Ohguri et al. [60]	NCI-CTCAE v3.0	Grade 3: 8 leukopenia, 7 neutropenia, 4 anaemia, 2 thrombocytopenia, 1 diarrhoea; grade 4: 1 neutropenia (= for HT cohort of 20)	None
Yamada et al. [54]	NR	NR	HT-related:1/17 suffered from severe subcutaneous fatty burn
Maluta et al. [62]	NCI-CTCAE v4.0	Grade 3: 5 diarrhoea	HT-related: 10/40 reported local discomfort which disappeared after rearrangement of treatment set-up
Ashayeri et al. [55]	NR	NR	NR
Maebayashi et al. [66]	NCI-CTCAE v4.0	Grade 3: 2 gastroenteritis; 2 haematotoxicity (= for HT cohort of 5).	None
Douwes [58]	NCI-CTCAE (version NR)	Grade 3: 2 leukopoenia, 3 anaemia	None
Tschoep-Lechner et al. [64]	NCI-CTCAE v4.0	Grade 3: 6 anaemia, 3 leukopoenia, 1 nausea, 1 febrile neutropenia; no HT-related toxicity	HT-related (grade 1–2): discomfort due to bolus pressure (3% of 201 HT treatments), position-related pain (17%), power-related pain (7%)
Volovat et al. [65]	NCI-CTCAE v2.0	ChT-releated: neutropenia grade 3 – 24%; anaemia grade 3 – 8%, thrombopenia grade III – 6%; neurologic toxicity grade 3 – 22%. (n = 19)	HT-related (grade 1–2): discomfort due to bolus pressure (2% of 229 HT treatments), position-related pain (12%), power-related pain (2%)
Ishikawa et al. [63]	NCI-CTCAE v3.0	Grade 3/4: 1 anorexia, 1 haemorrhage, GI-duodenum; 1 leucopoenia; 6 neutropenia; 3 anaemia	HT-related: all were mild and included pain and skin rash
Bakshandeh-Bath et al. [61]	WHO criteria	Cycle 1 (n = 13): none. Cycle 2 (n = 13): none. Cycle 3 (n = 4): grade 3: 1 neutropenia, 1 thrombocytopenia; 2 haematologic. Cycle 4 (n = 4): grade 3: 1 neutropenia; grade 4: 1 thrombocytopenia, 1 haematologic.	None
Koukoulias et al. [56]	WHO criteria	None	None
Kakehi et al. [53]	NR	NR	NR
Bull et al. [59]	NCI-CTCAE v3.0	NR for pancreas cohort	None

NCI-CTCAE: National Cancer Institute-Common Terminology Criteria for Adverse Events; NR: not reported; WHO: World Health Organisation.

Discussion

This is the first systematic review investigating the potential benefit of hyperthermia added to radiotherapy and/or chemotherapy for pancreatic cancer patients. In this review, in which 14 studies were included, we found for HT combined with chemo(radio)therapy a median OS of 6–18.6 months and a combined overall response rate of 31.3%. Results were consistent, showing longer median overall survival for the cohorts treated with HT compared to the respective control cohorts. Also response, reported for three of the studies that included a control cohort, was better with HT added. These results suggest that HT, when added to chemotherapy and/or radiotherapy, may positively affect response and overall survival for patients with pancreatic cancer. However, quality of the studies was limited and none of the studies was randomised. As a consequence, we did not perform a full meta-analysis of the data. Hyperthermia can safely be used, as reported toxicity was limited, with no grade 5 toxicity and only one case of a hyperthermia-related grade 3 burn.

Apart from the possible benefit in response and overall survival, the addition of hyperthermia may also contribute to a better quality of life (QoL). One controlled study included QoL measurements and showed that patients who received hyperthermia had a significantly larger improvement in QoL and a significant reduction in pain compared with those who did not receive hyperthermia [57]. Although these results are compelling, we did not include them in our analysis, as no other studies investigated the effects of hyperthermia on QoL. Including QoL as an endpoint in future studies will be highly valuable.

Thermal dosimetry

Knowing the actual temperature in the tumour during treatment is critical in HT, to verify that the therapeutically required temperature is obtained [69]. However, in only three studies, temperature was measured intratumorally and in six studies temperature was not measured at all. Frequently, heating is decreased when patients report pain. This may bring the temperature below the required temperature, as reported by Yamada et al. [54]; although their target temperature was 42.5 °C, only in five of the 17 patients the tumour was heated to more than the required 42 °C.

In this review, only one subcutaneous fatty burn was reported in one patient. Especially at the interface of muscle and fat, local hot spots can occur [69]. Thus, heating methods that allow for local adjustment of the temperature distribution are preferred, such as radiative heating using a phased array of multiple antennas [70]. Radiative heating also allows for HT treatment planning [71–74], which may yield better temperature distributions and the possibility to avoid hot spots. With temperature being heterogeneous and not constant in time, reporting is preferably done using a set of thermal dose parameters [69]. Such reporting allows for proper thermal dosimetry and enables (future) linking of thermal dose with outcome.

Limitations

The largest limitation of this review was the quality of the studies. All studies were cohort studies (prospective or retrospective); none of the studies were randomised controlled trials. Although the total number of included patients was moderate (n = 281), the individual studies were relatively small with the cohorts who received HT ranging from five to 40 patients. Only six out of 14 studies included a control group, but without randomisation, selection bias cannot be excluded. In addition, stratification according to tumour stage was not possible, as many studies included both patients with locally advanced and metastatic disease and did not report outcome per stage separately. The combining of locally advanced and metastatic disease may contribute to the large variation in outcome between studies.

Future work

The typically large stromal compartment of pancreatic cancer is often hypoxic, aiding tumour cells to escape conventional treatments. Hyperthermia can help to overcome the hypoxia, making those cancer cells more sensitive to radiotherapy and/or chemotherapy [12]. Interestingly, although hypoxia is thought to play a large role in the efficacy of HT, none of the studies included in our systematic review reported on the hypoxic state of the pancreatic tumours. This might in part be due to the fact that most patients in the included studies were treated over a decade ago, when non-invasive assessment of tumour hypoxia was in its infancy. Translational studies coupled to randomised trials that focus on the predictive value of hypoxia for efficacy of hyperthermia may supply further insight into the mechanisms behind HT.

Due to the small size and limited quality of the available clinical studies, the evidence for a potential advantage of adding HT to radiotherapy and/or chemotherapy in pancreatic cancer is supporting but weak. Currently, the Hyperthermia European Adjuvant Trial (HEAT) is being conducted, a randomised two-armed open study in patients with R0/R1 resected pancreatic carcinoma (NCT01077427; EudraCT No. 2008–004802-14). This study intends to include 336 patients in total, with patients in one arm treated with gemcitabine alone and in the other arm with gemcitabine plus cisplatin with regional HT. As of September 2017, 100 patients have been included.

Indeed, to establish the role of hyperthermia as an additional modality to the currently applied treatments for pancreatic cancer, large well-conducted randomised controlled trials are necessary. Reporting should be performed by intention to treat and should include overall survival, response, QoL, and tumour stage and toxicity reporting according to well-established criteria, to enable comparison between studies. Also proper thermometry [69], with accurate registration of the temperature throughout the HT treatment, is essential both for quality assurance of the delivered treatment as well as for linking treatment to clinical outcome.

Conclusions

This systematic review suggests that hyperthermia added to chemotherapy and/or radiotherapy may improve median OS and response rate of patients with locally advanced or metastatic pancreatic cancer. Randomised clinical trials are clearly needed to establish potential benefit of hyperthermia, with comprehensive reporting on stage, treatment including temperatures, response, duration of response, overall survival, quality of life and toxicity. Preferably, such trials will include translational studies that can help uncover the underlying mechanisms of the enhancement of chemotherapy and radiotherapy by hyperthermia.

Disclosure statement

The authors report no conflict of interest. The authors alone are responsible for the content and writing of the paper.

Appendix. Ovid search strategies

Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R)

Daily and Ovid MEDLINE(R) < 1946 to Present>.

Search date: 28 June 2017

Table

Step		n
1	hyperthermia, induced/	14,817
2	(hyperthermia or thermotherapy or thermal).ab, kf, ti.	178,091
3	hyperthermia.jw.	2269
4	or/1-3 [hyperthermia]	184,321
5	pancreatic neoplasms/or pancreas/	122,954
6	(Pancreatic or pancreas).ab, jw, kf, ti.	224,172
7	5 or 6	248,505
8	4 and 7	843
9	remove duplicates from 8	824

Embase Classic + Embase <1947 to 2017 June 26>. Ovid interface.

Search date: 28 June 2017

Table

Step		n
1	Thermotherapy/or exp hyperthermia/or onco hyperthermia device/	27,067
2	(Hyperthermia or thermotherapy or thermal).ab, kw, ti.	194,866
3	Hyperthermia.jx.	2372
4	(Hypertherm* or thermotron).dv.	45
5	or/1–4 [hyperthermia]	203,696
6	Pancreas tumour/or exp pancreas cancer/	103,352
7	(Pancreatic or pancreas).ab, jx, kw, ti.	309,358
8	6 or 7	332,029
9	5 and 8	1145
10	Remove duplicates from 9	1115

n: number of records..