Neuromodulation: Technology at the Neural Interface

Received: January 19, 2022 Revised: March 2, 2022 Accepted: March 6, 2022

https://doi.org/10.1016/j.neurom.2022.03.003

Effectiveness and Safety of Intrathecal Drug

Delivery Systems for the Management of Cancer
Pain: A Systematic Review and Meta-Analysis
Rui Duarte, PhD1 ; Sue Copley, MSc2; Sarah Nevitt, PhD1;
Michelle Maden, PhD1; Ali Mohammed Al-Ali, MD2,3; Denis Dupoiron, MD4;
Sam Eldabe, MD2
ABSTRACT
Objectives: Intrathecal drug delivery systems (IDDS) and spinal cord stimulation (SCS) have been proposed and assessed for the
management of cancer pain; however, such treatments remain underused. We conducted a systematic review to evaluate the
effectiveness and safety of IDDS and SCS for cancer pain.
Materials and Methods: Electronic databases MEDLINE, CENTRAL, EMBASE, and WikiStim were searched from 1988 to March
2021. Randomized controlled trials and observational studies of adults with pain related to cancer or its treatment who received
an implantable IDDS or SCS were eligible for inclusion. The primary outcome of the review was change in pain intensity from
baseline to the last available follow-up, measured using a visual analog scale or numerical rating scale. The protocol for this
review is registered on PROSPERO (CRD42021240717).
Results: A total of 22 studies (24 reports) included a total of 3043 participants who received either IDDS or SCS for cancer pain. Eight
studies reporting data for 405 participants with an IDDS could be included in the meta-analysis of pain intensity that showed a sta-
tistically significant reduction at the latest posttreatment follow-up time compared with baseline (mean difference [MD], −3.31; 95%
CI, −4.18 to −2.45; p < 0.001). Six studies reporting data for 325 participants with an IDDS could be included in the meta-analysis of pain
intensity that showed a statistically significant reduction up to one month after treatment compared with baseline (MD, −3.53; 95%
CI, −4.06 to −3.00; p < 0.001). A meta-analysis including studies of participants with either an IDDS or an SCS device showed similar
results. Improvements in other outcomes following implantation of IDDS also were observed. Postdural puncture headache was the
most reported complication, whereas urinary retention, nausea, and vomiting were commonly reported side effects.
Conclusion: Our findings suggest that IDDS is effective in reducing pain intensity for patients with cancer pain when compared
with pretreatment.

Keywords: Cancer pain, intrathecal drug delivery systems, meta-analysis, spinal cord stimulation, systematic review
Conflict of Interest: Rui Duarte, Sue Copley, and Sarah Nevitt have received consulting fees from Medtronic. Denis Dupoiron has
received consulting fees and speaker fees from Medtronic and Riemser. Sam Eldabe has received consulting fees and speaker
fees from Medtronic and Riemser. The other authors reported no conflict of interest.

INTRODUCTION to modern oncology treatments. An estimated 19.3 million new

cancer cases and almost 10.0 million cancer deaths occurred
The prevalence of cancer is increasing worldwide, as is the worldwide in 2020.1 It is expected that 28.4 million new cases will
number of patients who survive their cancer for many years thanks occur by 2040, a 47% rise from 2020.1 One of the most frequent

Address correspondence to: Rui Duarte, PhD, Liverpool Reviews and Implementation Group, University of Liverpool, Whelan Bldg, Liverpool L69 3GB, UK.
Email: [email protected]
1
Department of Health Data Science, Liverpool Reviews and Implementation Group, University of Liverpool, Liverpool, UK;
2
Department of Pain Medicine, The James Cook University Hospital, Middlesbrough, UK;
3
Jordan University of Science and Technology, Irbid, Jordan; and
4
Anesthesiology and Pain Department, Institut de Cancerologie de l’Ouest Paul Papin, Angers, France

For more information on author guidelines, an explanation of our peer review process, and conflict of interest informed consent policies, please see the journal’s Guide
for Authors.
1126

Source(s) of financial support: This study was funded by Medtronic Ltd. The funding source had no role in the study design, analysis, interpretation of data, writing of the
manuscript, approval, or decision to submit the manuscript for publication.

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 EFFECTIVENESS OF IDDS FOR CANCER PAIN

and disabling symptoms of cancer is pain,2 caused either by the search strategies are presented in Supplementary Data 1 of this
cancer or by its treatment. Severe pain in cancer survivors as a article. Search results were exported to an EndNote X9 (Clarivate,
result of the disease or treatments such as chemotherapy, radia- Philadelphia, PA) library and deduplicated. The reference lists of
tion, and surgery presents a significant challenge, given the het- relevant systematic reviews and eligible studies were hand-
erogeneous nature of pain presentation and individual experience.3 searched to identify further potentially relevant studies.
A systematic review and meta-analysis reported that 52% of
patients with cancer experienced pain, regardless of the stage of Study selection
their disease, and 33% continued to report pain even after curative The citations identified were assessed for inclusion in the review
treatment.4 Approximately 70% of patients achieve a good level of using a two-stage process. First, two reviewers (Rui Duarte and Sue
analgesia with appropriate medical management, but this leaves a Copley) independently screened all the titles and abstracts identi-
significant proportion of patients with unsatisfactory pain or fied by the electronic database searches to identify the potentially
symptom relief,5 which is associated with decreased quality of life relevant articles to be retrieved. Second, full-text copies of these
and with high societal and health care costs. studies were obtained and assessed independently by two
For patients with pain refractory to conventional medical man- reviewers (Sue Copley and Ali Mohammed Al-Ali) for inclusion
agement (CMM), interventional approaches may provide substan- using the eligibility criteria outlined in Table 1. Any disagreements
tial pain relief. These modalities include local anesthetic and steroid were resolved through discussion at each stage, and if necessary, in
injections, neuraxial analgesia (intrathecal or epidural catheter), and consultation with a third reviewer (Sam Eldabe or Rui Duarte).
sympathetic blockade from neurolytic injections and radio-
frequency or neurosurgical ablation.3 Spinal cord stimulation (SCS)
Data extraction
and intrathecal drug delivery systems (IDDS) have been proposed
A data extraction form was developed for the purposes of this
and evaluated as treatments for cancer-related pain.3 Despite IDDS
review. Data collected refers to study author and year of publica-
being routinely commissioned in England for the treatment of
tion, country where the study was conducted, funding sources,
severe cancer pain,6 there has not been a significant uptake in the
demographic data (ie, age, sex), intervention, comparator, duration
use of IDDS since the publication of the clinical commissioning
of follow-up, type of cancer, type of pain, duration of pain because
policy by the National Health Service England.7 Several reviews
of cancer, months since diagnosis, number of participants included
have been performed to investigate either the effectiveness or the
in the analysis, and outcomes. Data extraction was performed by
safety, or both, of IDDS,8,9 SCS,3,10 or IDDS and SCS11,12 for cancer
one reviewer (Sue Copley) and checked for accuracy by two
pain. However, some of the aforementioned were narrative
reviewers (Rui Duarte and Sarah Nevitt).
reviews; furthermore, several reports have been published in recent
years, and no meta-analysis has previously been performed.
The aim of this systematic review is to evaluate the effectiveness Risk of bias assessment
and safety of IDDS and SCS for the management of cancer pain. Risk of bias (RoB) of included randomized controlled trials (RCTs)
was assessed by using the Cochrane RoB tool (RoB 2.0).15 The
National Institutes of Health study quality assessment tools16 for
MATERIALS AND METHODS cohort studies, case-control studies, and before-after (pre-post)
studies with no control group were used to assess the RoB of
The systematic review methods followed the general principles included nonrandomized studies. RoB assessment of the included
outlined in the Centre for Reviews and Dissemination (CRD) guid- studies was undertaken by one reviewer (Rui Duarte) and checked
ance for conducting reviews in health care.13 This systematic review by a second reviewer (Sarah Nevitt). Any disagreements were
is reported in accordance with the Preferred Reporting Items for resolved by discussion and, if necessary, in consultation with a third
Systematic Reviews and Meta-Analyses (PRISMA).14 The protocol for reviewer (Sam Eldabe).
this review is registered on PROSPERO (CRD42021240717).
Outcomes
Search strategy The primary outcome of the review was change in pain intensity
Electronic databases MEDLINE, CENTRAL, EMBASE, and WikiStim from baseline to the last available follow-up, measured using a
were searched by an information specialist (Michelle Maden) from visual analog scale (VAS) or numerical rating scale (NRS). Additional
1988 (introduction of IDDS into clinical practice for management of outcomes included change in pain intensity from baseline up to
cancer pain) to March 24, 2021, without language restrictions. The one month after treatment, proportion of patients achieving at

Table 1. Eligibility Criteria.

Inclusion criteria (if all of the following met) Exclusion criteria (if any of the following met)
1. Population comprised adult patients (18 y or older) with pain related to 1. IDDS or SCS used for management of noncancer conditions
cancer or its treatment
2. Intervention was implantable IDDS or SCS or IDDS and SCS used 2. Neuromodulation intervention other than IDDS or SCS
concurrently
3. Any comparator or no comparator 3. Case reports, design/protocol paper, methodological paper, (systematic)
review, meta-analysis, commentaries/editorial
4. RCTs, case-control studies, cohort studies, uncontrolled single-arm studies, 4. Insufficient information (eg, study only available as a conference
1127

and registry studies with sample size ≥ 20 participants proceeding/abstract)

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 DUARTE ET AL

least a 50% reduction in pain from baseline to the last available Development and Evaluations (GRADE) framework.32 The magni-
follow-up, change in health-related quality of life (HRQoL) from tude of effect and certainty of the evidence were considered when
baseline to the last available follow-up, measured using a validated drawing conclusions.
HRQoL questionnaire (eg, EQ-5D), change in functional outcomes For other outcomes (proportion of patients achieving at least a
from baseline to the last available follow-up (eg, Oswestry Disability 50% reduction in pain from baseline, HRQoL, functional outcomes,
questionnaire), reduction in systemic opioid intake, survival, device- opioid intake, survival, and the proportion of patients with com-
related complications, and side effects. plications or side effects), we deemed that clinical heterogeneity
was too great, or outcome data reported were too variable or too
Data synthesis limited, to allow data to be pooled in the meta-analysis. A narrative
Clinical heterogeneity of included studies was assessed by synthesis of these outcomes was performed.
comparing study design characteristics, participant characteristics, Data for other outcomes were too limited and heterogeneous to
and definitions of outcomes. formally apply the GRADE framework to assess the certainty of the
Only one included study17 included a control group of partici- evidence. The effect of IDDS and SCS on these outcomes in adults
pants who did not receive IDDS or SCS. Therefore, it was not with cancer pain is very uncertain.
possible to perform a meta-analysis for the comparison of IDDS or
SCS vs control. Comparative results from an RCT by Smith et al17
and follow-up studies18,19 were reported narratively. RESULTS
We deemed studies to be sufficiently clinically homogeneous in Study Selection
terms of study characteristics and population (Tables 1 and 2), with After deduplication, our database searches identified a total of
sufficient outcome data reported to perform a meta-analysis for the 2636 records. Following the initial screening of titles and abstracts,
outcome mean difference (MD) in pain intensity after treatment 69 records were considered to be potentially relevant and were
compared with baseline. Random-effects meta-analysis was per- retrieved to allow assessment of the full-text publication. After a
formed using the generic-inverse variance method, using the metaan review of the full-text publications, a total of 22 studies (24
command27 in Stata version 14.0 (StataCorp, College Station, TX), with reports) were judged to meet the inclusion criteria for this
restricted maximum likelihood used to estimate heterogeneity vari- review.17-26,29,31,33-44 Citations and reasons why studies were
ance.28 The meta-analysis was stratified by prospective or retrospec- excluded on the review of the full publication are listed in
tive study design, and the meta-analysis was conducted separately for Supplementary Data 3. The PRISMA flow diagram summarizing the
up to one month after treatment and at the latest posttreatment study selection process is presented in Figure 1.
follow-up time reported. We assessed the statistical heterogeneity in
the meta-analysis according to the I2 statistic, with higher values
corresponding to higher levels of statistical heterogeneity. Characteristics of included studies
To standardize the outcome data to a single scale, we assumed that The included studies recruited a total of 3043 people who
the VAS (0–10 cm) and the NRS (0–10) were equivalent, and we con- received either IDDS or SCS for the management of cancer pain.
verted the VAS (0–100 mm) by dividing pain scores by ten. One study29 The characteristics of the included RCTs and prospective studies are
reported individual participant data for the 31 included participants, presented in Table 2. The characteristics of the included retro-
and we used these data to estimate a within-participant correlation spective studies are presented in Table 3. One RCT was identified
value of pain intensity scores at baseline and four to six weeks after that compared IDDS with CMM.17 Two reports provided a post hoc
treatment of 0.383. Using this within-participant correlation value, we analysis of this RCT.18,19 One of the post hoc reports evaluated 30
were able to calculate MD in the change from baseline and the stan- patients who crossed over to IDDS after not obtaining satisfactory
dard error of the mean difference (SE [MD]) for all studies included in pain relief with CMM (defined as pain score on a VAS persistently
the meta-analysis.30 Another study26 reported baseline, 12 months >5 despite maximum tolerated drug dosages).19 The other post
posttreatment, and change from baseline pain intensity scores, which hoc study compared those patients who received IDDS (patients
allowed us to estimate a within-participant correlation value of pain randomized to IDDS combined with those who crossed over from
intensity scores at baseline and at 12 months after treatment of 0.810. CMM) to patients who did not receive IDDS (patients randomized
We repeated the meta-analysis using this correlation value to calculate to CMM who did not cross over to IDDS combined with those who
the MD and SE [MD]. Numerical results of the meta-analysis were were randomized to IDDS but did not receive an implant).18 A total
similar, and conclusions were unchanged (Supplementary Data 2, of 21 studies were observational cohort studies, of which seven
Supplementary Data Figs. S1 and S2). were prospective20-26 and 14 were retrospective.29,31,33-44 Two
Of the nine studies included in quantitative synthesis, six studies studies included participants who received either IDDS or SCS for
reported average pain scores only17,21,23,26,29,31; one study reported the management of cancer pain.33,36 The remaining studies eval-
maximum pain scores only20; and two studies reported maximum uated IDDS only. All studies that provide information on the
pain scores, as well as average21 or minimum24 pain scores. We intrathecal medications administered report the use of an opioid
included the maximum pain scores or average pain scores (where (most commonly morphine) either alone or in combination with
maximum pain scores were not reported) in the meta-analysis, and adjuvants such as bupivacaine, ropivacaine, clonidine, and baclo-
we conducted a sensitivity analysis using the average or minimum fen. The use of ziconotide either alone or in combination was
pain scores (rather than the maximum pain scores) for the two reported in six studies.20-22,24,35,42 The follow-up duration ranged
studies reporting multiple pain scores. The numerical results of the from seven days to 16 months. The type of pain (ie, whether
meta-analysis were similar, and conclusions were unchanged neuropathic, nociceptive, or mixed) was not regularly reported.
(Supplementary Data 2, Supplementary Data Figs. S3 and S4). Similarly, the duration of pain because of cancer or time since
1128

We assessed the certainty of the evidence for the primary cancer diagnosis until IDDS or SCS implant was not commonly
outcome using the Grading of Recommendations, Assessment, reported.

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 www.neuromodulationjournal.org
Table 2. Characteristics of the Included Randomized and Prospective Studies.

Study characteristics Population

†
Author (country); funding Intervention Comparator Follow-up No. of patients in the Type(s) of cancer (%) Type(s) of pain (%) Duration of pain because
duration analysis, sex, and mean of cancer (i) or months
age ± SD* since diagnosis (ii)
(mean ± SD)*
RCTs
Smith et al17 (United IDDS CMM Primary end IDDS n = 101 (M = 52; IDDS—lung (19.8), breast IDDS—neuropathic (12.9), IDDS—(ii) median
States, Italy, Spain, IT opioids‡ point at F = 49), 56.2 ± 13.2 y (8.9), prostate (5.9), nociceptive (25.7), 17.7 mo (IQR 8–46.4)
Switzerland, and Morphine (48/51) 4 wk CMM n = 99 (M = 59; colon (5.0), pancreas mixed (61.4) CMM—(ii) median
Australia); funding: in Hydromorphone (3/51) Survival F = 40), 57.8 ± 13.7 y (6.9) CMM—neuropathic 18.6 mo (IQR 6.1–43.8)
part by Medtronic IT adjuvants at 6 mo CMM—lung (25.5), (14.3), nociceptive
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
International Neuromodulation Society. This is an open access article

Bupivacaine (15/51) breast (9.2), prostate (25.5), mixed (60.2)

© 2022 The Authors. Published by Elsevier Inc. on behalf of the

Clonidine (2/51) (11.2), colon (9.2),

Droperidol (1/51) pancreas (5.1)
Post hoc studies of Smith et al17 RCT
Smith and Coyne19 IDDS after failing CMM NA Primary end n = 30 (M = 16; F = 14), Lung (16.7), colon (13.3), Neuropathic (13.3), noci- (ii) Median 17.6 mo
(country and funding as point at 56.0 ± 13.4 y breast (10.0), prostate ceptive (16.7), mixed (IQR 5.1–35.6)
in Smith et al17) 4 wk (6.7), pancreas (6.7) (70.0)
Survival
at 6 mo
Smith et al18 (country and IDDS (IDDS implanted Non-IDDS (IDDS 4 and 12 wk IDDS n = 103 (M = 55; IDDS—lung (18.4), breast IDDS—neuropathic (14.6), –
funding as in Smith group + CMM implan- nonimplanted Survival F = 48) (9.7), prostate (7.8), nociceptive (22.3),
et al17) ted group) group + CMM at 6 mo Non-IDDS n = 97 colon (7.8), pancreas mixed (63.1)
nonimplanted (M = 56; F = 41) (5.8) Non-IDDS—
group) Non-IDDS—lung (26.8), neuropathic (12.4),
prostate (9.3), breast nociceptive (28.9),
(8.2), colon (6.2), mixed (58.7)
pancreas (4.1)
Prospective studies
Brogan et al20 (United IDDS NA 41.7 ± 25.7 n = 58 (M = 40; F = 18), Prostate (13.8), colorectal NR NR
States); Funding: no IT opioids d (range 56.8 ± 14.6 y (10.3), pancreas (8.6),
financial support Morphine or 14–82) breast (6.9), lung (6.9)
hydromorphone (53/

EFFECTIVENESS OF IDDS FOR CANCER PAIN

58)
IT opioids alone (23/58)
IT adjuvants
Neuromodulation 2023; 26: 1126–1141

Bupivacaine (24/58)
Ziconotide (9/58)
Ziconotide alone (4/58)
(Continued)

1129
 1130
www.neuromodulationjournal.org

DUARTE ET AL
Table 2. Continued

Study characteristics Population

†
Author (country); funding Intervention Comparator Follow-up No. of patients in the Type(s) of cancer (%) Type(s) of pain (%) Duration of pain because
duration analysis, sex, and mean of cancer (i) or months
age ± SD* since diagnosis (ii)
(mean ± SD)*

Brogan et al21 (United IDDS NA 8 wk n = 51 (33; F = 18), Prostate (17.6), breast NR NR

States); Funding: IT opioids§ 56.6 ± 14.3 y (13.7), pancreatic (11.8),
Department of Morphine (31/36) lung (11.8), sarcoma
Anesthesiology, Hydromorphone (5/36) (9.8)
University of Utah IT opioids alone (10/36)
IT adjuvants
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
International Neuromodulation Society. This is an open access article

Bupivacaine (24/36)
© 2022 The Authors. Published by Elsevier Inc. on behalf of the

Clonidine (1/36)
Ziconotide (4/36)
Carvajal et al22 (France); IDDS NA 3 mo n = 70 (M = 39; F = 31), Pancreatic (100) NR (ii) Median 8.3 mo (IQR
Funding: no financial IT opioids median 62.5 y (IQR 1.8–13.9)
support Morphine (70/70) 51.5–68)
IT adjuvants
Ropivacaine, clonidine,
or ziconotide provided
as adjuvants (n specific
to implantable IDDS
unclear)
Cheng et al23 (China); IDDS NA 12 mo n = 33 (M = 21; F = 12), Lung (27.3), gastrointes- NR (i) 0.58 ± 0.38 y
Funding: Jiangsu Pro- Morphine (33/33) 58.6 ± 7.0 y tinal (21.2), hep-
vincial Medical Youth atobiliary (15.2),
Talents Program and pancreatic (12.1), breast
the National Natural (9.1)
Science Foundation of
China
Dupoiron et al24 (France); IDDS NA 3 mo n = 97 (M = 52; F = 45), Pancreatic, colorectal, NR NR
Funding: no financial Morphine, ropivacaine, 59.8 y (range 25–88) prostate, breast, uterus
support clonidine, ziconotide
Rauck et al25 (United IDDS NA 16 mo n = 119 (M = 60; F = 59), Lung (26.3), breast (16.9), NR NR
States and international Morphine 60.6 ± 13.5 y colorectal (13.6), pros-
sites); Funding: tate (9.3)
Neuromodulation 2023; 26: 1126–1141

Medtronic
Stearns et al26 (United IDDS NA 12 mo n = 1403 (M = 609; F = Lung (15.0), breast (11.0), NR (ii) Median 28.9 mo (IQR
States, Europe, and 794), 59 y (range 13–93) colon/rectal (10.8), 14.9–66.0)
Latin America); Fund- pancreatic (8.3), pros-
ing: Medtronic tate (5.9)
F, female; IQR, interquartile range; IT, intrathecal; M, male; NA, not applicable; NR, not reported.
*Age, duration of pain because of cancer, and months since diagnosis reported as mean ± SD unless otherwise stated.
†
Five most reported types of cancer.
‡
Denominator refers to the number of patients who went on to receive an IDDS.
§
Information on IDDS solution only presented for a subset of the population.
 EFFECTIVENESS OF IDDS FOR CANCER PAIN

Figure 1. PRISMA 2020 flow diagram. [Color figure can be viewed at www.neuromodulationjournal.org]

RoB assessment as such, item 4 was rated as CD. However, because most reports
The full RoB assessment for each included study is presented in were of single-center studies, it should be considered that the
Supplementary Data 4. The included RCT17 was judged to have a intervention would have been delivered consistently to all patients.
low RoB for the domains of the process of randomization and level Some studies provide a description of the initial drug selection,
of missing outcome data. The domain deviation from intended starting dose, and escalation.20-22,24,35,42 Only one of the non-
interventions was judged as presenting some concerns because of randomized studies provided a statistical rationale for their sample
the number of patients allocated to IDDS who did not receive the size (item 6).21 None of the nonrandomized studies mentioned
assigned intervention. The RCT was judged to have a high RoB for whether the outcome assessors were blinded to the intervention
outcome measurement because it was an open-label trial and (item 8). Although it would not be possible to blind patients, cli-
because of the plausibility that knowledge of the treatment nicians, or nurses responsible for the care of the patients to the use
received and beliefs about potential beneficial effects could have of SCS or IDDS, an investigator collecting the outcomes unaware of
influenced the outcomes. The domain selection of the reported the intervention could be considered as blinded. This would only
result was judged as presenting some concerns because there was be applicable to prospective data collection and not retrospective
no mention whether the statistical analyses followed a prespecified studies involving reviews of medical records. Although several of
statistical analysis plan. The overall bias for the included RCT was the studies had loss to follow-up of at least 20% (item 9), it should
considered to be high. be considered that such an attrition rate would be expected in a
Most of the nonrandomized studies were conducted in a single population of patients with cancer pain, particularly in those
center or single country, and as such, it is not possible to determine studies with longer follow-up time points.
whether the patients included would be representative of other
patients with cancer pain managed with IDDS. Even in a large
registry study, most patients (81%) were from a single center.26 Outcomes of included studies
Therefore, item 3 on representativeness of the sample was rated The outcomes of the included RCTs and prospective studies are
as “cannot determine” (CD) for all studies. In many instances, the presented in Table 4, and the outcomes of the included retro-
IDDS procedure was poorly described in the included studies, and spective studies are presented in Table 5.
1131

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 1132
www.neuromodulationjournal.org

DUARTE ET AL
Table 3. Characteristics of the Included Retrospective Studies.

Study characteristics Population

Author (country); funding Intervention Comparator Follow-up duration No. of patients in the Type(s) of cancer (%)† Type(s) of Duration of pain because
analysis, sex, and mean pain (%) of cancer (i) or months
age ± SD* since diagnosis (ii)
(mean ± SD)*
Retrospective studies
Aigner et al33 (United IDDS or SCS NA FU1 median 1.5 mo IDDS n = 49/SCS n = 30 Colorectal (13.9), lung NR NR
States); funding: Cancer FU2 median 3.4 mo (M = 37; F = 42), (10.1), sarcoma (10.1),
Prevention Fellowship, FU3 median 7.2 mo 54 ± 14.3 y breast (8.9), renal (8.9)
National Cancer Insti-
tute and National Insti-
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
International Neuromodulation Society. This is an open access article

tutes of Health
© 2022 The Authors. Published by Elsevier Inc. on behalf of the

Becker et al34 (Germany); IDDS NA Neuropathic pain median n = 43 (M = 24; F = 19), 64 Lower gastrointestinal Neuropathic (i) Median 11 mo
funding: NR 2.5 mo y (range 40–84) tract (37.2), lung (11.6), (46.5),
Nociceptive pain breast (11.6), prostate nociceptive
median 5 mo gland (9.3), urinary tract (53.5)
(9.3)
Brogan and Winter29 IDDS NA 4 to 6 wk n = 31 (M = 20; F = 11), Colorectal (22.6), NR (ii) Mean 30.4 mo (IQR
(United States); funding: IT opioids 55.2 ± 12.5 y pancreatic (19.4), 7.5–40)
no financial support Morphine or sarcoma (12.9), prostate
hydromorphone (9.7), lung (6.5)
(31/31) IT adjuvants
Bupivacaine (28/31)
Clonidine (5/31)
Baclofen (2/31)
Chen et al31 (United IDDS NA Mean 8.9 d, median 4 d n = 43 (M = 19; F = 24), Pancreatic (13.9), breast NR NR
States); funding: MSK Morphine or median 59 y (13.9), renal (9.3),
Cancer Center Support hydromorphone with ovarian (9.3), sarcoma
Grant and Department bupivacaine (6.9)
of Anesthesiology and
Critical Care
Dupoiron et al35 (France); IDDS NA 7d n = 108 (M = 59; F = 49), Digestive tract (35.0), Neuropathic (51) NR
funding: NR IT opioids 61 ± 11 y urologic tract (20.5), Nonneuropathic
Morphine (108/108) lung (19.1), gynecologic (49)
IT adjuvants tract (16.8), other (8.6)
Ropivacaine (108/108)
Neuromodulation 2023; 26: 1126–1141

Ziconotide (108/108)
Engle et al36 (United IDDS or SCS NA 12 mo n = 131 (M = 69; F = 62), NR NR NR
States); funding: NR median 53 y
Erdine et al37 (Turkey); IDDS NA 235.4 ± 140.4 d (range n = 54 (M = 39; F = 15), Lung (50), gastrointestinal NR NR
funding: NR Morphine (54/54) 70–727) 56.0 ± 11.3 y (27.8), urogenital (7.4),
breast (7.4), others (7.4)
Onofrio and Yaksh38 IDDS NA 4 mo n = 53 (M = 24; F = 29), Colon (16.9), pancreas NR (i) 16.1 ± 4.1 mo (range
(United States); funding: Morphine (53/53) 58.4 ± 1.6 y (15.1), rectal (11.3), 1–216)
NR cervical (11.3), lung (9.4)
(Continued)
 www.neuromodulationjournal.org
Table 3. Continued

Study characteristics Population

Author (country); funding Intervention Comparator Follow-up duration No. of patients in the Type(s) of cancer (%)† Type(s) of Duration of pain because
analysis, sex, and mean pain (%) of cancer (i) or months
age ± SD* since diagnosis (ii)
(mean ± SD)*

Reig and Abejón39 (Spain); IDDS NA Median 6.37 mo (IQR n = 64 (M = 34; F = 30), NR NR NR
funding: NR Morphine (64/64) 1–59) median 57.5 y (IQR
A proportion of 16–77)
patients will have
received IT adjuvants
Sayed et al40 (United IDDS NA 12 mo n = 160 (M = 83; F = 77), Pancreatic (20) NR NR
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
International Neuromodulation Society. This is an open access article

States); funding: no IT opioids and bupiva- median 58 (IQR 18–79)

© 2022 The Authors. Published by Elsevier Inc. on behalf of the

financial support caine in a proportion of

patients as IT adjuvant
Scanlon et al41 (United IDDS NA 12 mo n = 64 (M = 29; F = 35), 60 NR NR NR
States); funding: NR y
Sindt et al42 (United IDDS NA 30 d n = 173 (M = 96; F = 77), Pancreas (13.3), breast NR NR
States); funding: no Opioid alone (38/173) 54.5 ± 15.3 y (11.0), prostate (9.8),
financial support Opioid + bupivacaine lung (9.2), other (56.6)
(110/173)
Opioid + ziconotide
(11/173)
Three or more medica-
tions (9/173)
Ziconotide alone
(3/173)
Ziconotide + bupiva-
caine (2/173)
Spiegel et al43 (United IDDS NA 6 mo n = 50 (M = 22; F = 28), Gastrointestinal and NR NR
States); funding: Morphine or hydro- median 61 y breast most common
Memorial Sloan Ketter- morphone in combina- cancer types
ing Cancer Center Sup- tion with bupivacaine
port Grant and the

EFFECTIVENESS OF IDDS FOR CANCER PAIN

Department of Anes-
thesiology and Critical
Care
Neuromodulation 2023; 26: 1126–1141

Yegul et al44 (Turkey); IDDS NA Mean 3.2 mo n = 58 Not specified for IDDS NR NR
funding: NR patients separately
F, female; FU, follow-up; IQR, interquartile range; IT, intrathecal; M, male; NA, not applicable; NR, not reported.
*Age, duration of pain because of cancer, and months since diagnosis reported as mean ± SD unless otherwise stated.
†
Five most reported types of cancer.

1133
 1134
www.neuromodulationjournal.org

DUARTE ET AL
Table 4. Outcomes of the Included Randomized and Prospective Studies.

Author Pain intensity Functional, HRQoL outcomes and Systemic opioid intake Complications/side effects
survival
RCTs
Smith IDDS (n = 71) VAS (mean ± SD): baseline Cumulative survival was 53.9% at 6 mo IDDS IDDS (n) CMM (n)
et al17 7.57 ± 1.79/4-wk 3.90 ± 3.42 CMM compared with 37.2% for the CMM Baseline median daily MOED 250 mg/4-wk Patients with SAE 62 69
(n = 72) VAS (mean ± SD): baseline group (p < 0.06 log-rank test) 50 mg CMM IDDS-related SAE 14 2
7.81 ± 1.63/4-w 3.05 ± 3.16 Baseline median daily MOED 272 mg/4-wk Pocket problems 4 2
290 mg Infections 1 1
Pump migration 1 –
Lumbar site 5 –
Catheter problems 5 –
CSF leak 1 –
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
International Neuromodulation Society. This is an open access article

Post hoc studies of Smith et al17 RCT

© 2022 The Authors. Published by Elsevier Inc. on behalf of the

Smith and VAS (mean ± SD): baseline 7.69 ± 2.22/4-wk Cumulative survival was 44% at 6 mo Baseline median MOED 320 mg (IQR –
Coyne19 4.5 ± 2.7 Median survival after implant 103 d 120–1240)
↑ p < 0.011
Smith Pain scores significantly better with IDDS Survival at 6 mo: – –
et al18 when compared with non-IDDS at 4-wk IDDS no implant 59.2%
(p = 0.002) but not at 12-wk (p = 0.23) IDDS implant 54.3%
CMM implant 51.8%
CMM no implant 31.5%
Prospective studies
Brogan NRS (mean ± SD): worst pain baseline MDASI SS (mean ± SD): baseline Baseline mean ± SD daily MOED 805.28 ± Based on 98 pumps implanted, rather than the
et al20 (n = 57) 8.32 ± 1.73/FU (n = 53) 4.98 ± 1.67/FU 3.72 ± 1.80 1085.81 mg (range 0–5760)/FU 128.18 ± 58 in the final study group. One pump infec-
4.98 ± 2.92 ↑ p < 0.0001 387.55 (range 0–2160) tion requiring explant and intravenous anti-
↑ p < 0.001 MDASI SI (mean ± SD): baseline ↑ p < 0.001 biotic therapy. Three patients (3.1%)
≥50% pain reduction = 44% 6.53 ± 2.22/FU 4.37 ± 2.54 84% reduction in daily MOED developed postdural puncture headache.
↑ p < 0.001 Several patients developed mild, transient,
lower-extremity weakness after PCIA dosing.
Urinary hesitancy was reported in several
patients
Brogan NRS (mean ± SD): worst pain baseline MDASI SS (mean ± SD): baseline (n = 51) Baseline mean daily MOED 375 mg (range No infectious complications during the study
et al21 (n = 51) 8.00 ± 1.80/4-wk (n = 32) 5.13 ± 1.90/4-wk (n = 32) 3.70 ± 2.13/8-wk 0–3160) period. Two patients developed a postdural
5.70 ± 2.58/8-wk (n = 24) 6.67 ± 2.73 (n = 24) 4.29 ± 1.89 At 4-wk 28/32 (87.5%) discontinued all puncture headache who did not respond to
NRS (mean ± SD): average pain baseline MDASI SI (mean ± SD): baseline (n = 51) non-IT opioids; at 8-wk 22/24 (92%) conservative therapy and required an epidural
(n = 51) 5.86 ± 1.8/4-wk (n = 32) 5.88 ± 2.63/4-wk (n = 32) 3.92 ± 2.82/8-wk remained off all non-IT opioids blood patch
4.00 ± 2.44/8-wk (n = 24) 4.54 ± 2.72 (n = 24) 4.22 ± 3.00
Neuromodulation 2023; 26: 1126–1141

MDASI fatigue score (mean ± SD): baseline

(n = 51) 7.84 ± 1.90/4-wk (n = 32)
5.78 ± 2.80/8-wk (n = 24) 6.96 ± 2.76
Carvajal NRS baseline median 8 (IQR 7–9)/difference Median OS 91 d (95% CI 83–111) Baseline median daily MOED 385.5 mg Postlumbar puncture headache 30 (32.3)
et al22 from baseline at 1-wk (n = 69) −6 (−7 (range 265–600) Catheter/pump malfunction 3 (3.2)
to −4)/4-wk (n = 66) −5 (−6 to −4)/3 mo Surgical wound dehiscence 2 (2.1)
(n = 32) −5.5 (−7 to −4) Pump pocket infection 1 (1.1)
≥50% pain reduction at 1-wk 59/69 Bacterial meningitis 1 (1.1)
(85.5%)/4-wk 52/66 (78.8%)/3 mo 25/32 Subdural hematoma 1 (1.1)
(78.1%) Minor spinal cord puncture 1 (1.1)
Pocket hematoma 1 (1.1)
Pulmonary embolism 1 (1.1)
(Continued)
 www.neuromodulationjournal.org
Table 4. Continued

Author Pain intensity Functional, HRQoL outcomes and Systemic opioid intake Complications/side effects
survival

Cheng NRS (mean ± SD): baseline 7.76 ± 1.15* KPS baseline mean ± SD 72.27 ± 10.66 – Nausea and vomiting 11 (33.3)
et al23 SAS baseline mean ± SD 57.91 ± 11.82 Drowsiness 8 (24.2)
SDS baseline mean ± SD 62.12 ± 11.64 Constipation 5 (15.2)
Eight patients died within 12-mo FU Dizziness 4 (12.1)
Skin reactions 3 (9.1)
Hypotension 2 (6.1)
Diarrhea 1 (3.0)
Dupoiron NRS baseline (mean ± SE) 7.97 ± 0.27/4-wk – Baseline mean daily MOED 567 mg, reduced Postdural puncture headache 52 (54)
et al24 3.65 ± 0.46† to a mean daily dose of 5.35 mg following Urinary retention 16 (16)
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

treatment Withdrawal syndrome 38 (39)

International Neuromodulation Society. This is an open access article

Infection 5 (5)
© 2022 The Authors. Published by Elsevier Inc. on behalf of the

Hematoma 3 (3)
Rotation of pump in site 3 (3)
Rauck NAS (mean) decreased from 6.1 at baseline – Baseline mean ± SD daily MOED 106.5 ± Total of 63 SAEs in 40 patients; seven device-
et al25 to 4.2 at 1 mo (↑ p < 0.01) and remained 135.3 mg related SAEs in seven patients; 55 procedure-
decreased through month 13 (↑ p < 0.05) At 3, 4, and 13 mo more than half of related SAEs in 36 patients; and one SAE
≥50% pain reduction at 1 mo (n = 99) patients reported no use of systemic opi- related to intercurrent illness or injury in one
37%/2 mo (n = 74) 42%/3 mo (n = 59) oids. At each visit through 13 mo more patient
42%/4 mo (n = 47) (43%) than 70% of patients reported a reduction
of 50% or more in systemic opioid use
from baseline (↑ p < 0.01)
Stearns NRS (mean ± SD): baseline (n = 283) EQ-5D index score (mean ± SD): baseline – Total of 68 SAEs in 54/706 patients (7.7); 17
et al26 6.8 ± 2.4/6 mo (n = 103) 5.5 ± 2.6/12 (n = 139) 0.386 ± 0.252/6 mo (n = 41) infections and infestation events in 17 patients
mo (n = 55) 5.4 ± 2.5 (↑ p < 0.01) 0.556 ± 0.252 (2.4); five psychiatric disorder events in four
↑ p < 0.01 at 6 mo/(-) at 12 mo patients (0.6); nine nervous system disorder
Postimplant survival was 39%, 24%, 16%, events in nine patients (1.3); 29 general disor-
11%, and 5% at 0.5, 1, 2, 3, and 10 y, ders and administration site conditions in 23
respectively patients (3.3); six injury, poisoning and proce-
dural complications in six patients (0.9); two
respiratory, thoracic, and mediastinal disorders
in two patients (0.28)

EFFECTIVENESS OF IDDS FOR CANCER PAIN

279/706 (40) experienced ≥1 AE related to the
device components, implant procedure, or
delivery of therapy. Most frequently occurring
AEs were adverse drug reaction (24.5%; 95% CI,
Neuromodulation 2023; 26: 1126–1141

21.5–27.8) and medical device site pain (10.1%;

95% CI, 8.1–12.5)
(-) No statistically significant differences between assessments. ↑ Statistically significant improvements after IDDS.
AE, adverse event; CSF, cerebrospinal fluid; FU, follow-up; IQR, interquartile range; IT, intrathecal; KPS, Karnofsky performance scores; MDASI, MD Anderson Symptom Inventory; NAS, numerical analog scale;
OS, overall survival; PCIA, patient-controlled intrathecal analgesia; SAE, serious adverse event; SAS, self-rating anxiety scale; SDS, self-rating depression scale; SI, symptom interference with quality of life; SS,
symptom severity.
*FU data from Cheng et al23 only presented graphically for all outcomes and separated in two groups both receiving IDDS; pain intensity data were digitized and combined for the purposes of meta-
analysis.
†
Minimum and maximum pain intensity data at baseline, 30 d, 60 d, and 90 d were presented graphically in Dupoiron et al.24 Data were digitized for the purposes of meta-analysis.

1135
 1136
www.neuromodulationjournal.org

DUARTE ET AL
Table 5. Outcomes of the Included Retrospective Studies.

Author Pain intensity Functional, HRQoL outcomes and survival Systemic opioid intake Complications/side effects
Retrospective studies
Aigner et al33 ESAS (mean ± SD): baseline (n = 79) 5.9 ± Anxiety using ESAS (mean ± SD): baseline – Procedure related 13 (16.6)
2.07/FU1 (n = 74) 4.64 ± 2.16/FU2 (n = 47) (n = 74) 3.26 ± 2.81/FU1 (n = 70) 3.28 ± Infection 3 (3.8)
4.61 ± 2.08)/FU3 (n = 42) 4.84 ± 2.20 3.11/FU2 (n = 46) 2.83 ± 3.09)/FU3 (n = Confusion and nausea 3 (3.8)
42) 3.90 ± 3.07 Hematoma 1 (1.3)
Depression using ESAS (mean ± SD): Surgery revision 6 (7.6)
baseline (n = 75) 2.89 ± 3.00/FU1 (n =
69) 2.80 ± 2.94/FU2 (n = 47) 2.68 ±
2.87)/FU3 (n = 42) 2.90 ± 3.12
Becker et al34 Neuropathic VRS baseline (n = 20) median 9/ – – Transient side effects 23 (53.5)
FU 2.5 mo (n = 13) 8 Nausea and vomiting 15 (34.9)
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
International Neuromodulation Society. This is an open access article

Nociceptive VRS baseline (n = 23) median Procedure related 5 (11.6)

© 2022 The Authors. Published by Elsevier Inc. on behalf of the

9/FU 5 mo (n = 13) 3 Spinal catheter malfunction 3 (6.9)

Hematoma 1 (2.3)
Postoperative pneumonia 1 (2.3)
Brogan and Winter29 NRS (mean ± SD): baseline (n = 29) 6.5 ± 2.1/ Mean ± SD survival 3.5 ± 1.9 mo Baseline mean daily MOED 796 mg (range One patient had an increase in pain after
FU (n = 29) 3.1 ± 2.0 0–4.320)/FU 64 mg (range 0–803.8) the catheter had backed out of the
↑ p < 0.001* ↑ p < 0.001 intrathecal space. Urinary hesitancy
92% reduction in daily MOED was reported in several patients.
Chen et al31 VAS (mean ± SD): baseline 6.5 ± 2.0 (range – Baseline mean ± SD daily MOED 1041 ± 1267 Numbness and urinary retention 7 (16.2)
0–10)/FU 2.7 ± 2.3 (range 0–8) mg (range 24–5560)/FU 307 ± 543 mg Postdural puncture headache 3 (6.9)
(range 0–2545)
71% reduction in daily MOED
Dupoiron et al35 Minimum NRS baseline median 5 (range 0–8) Median overall survival 142 d (range Baseline median daily MOED 300 mg (range –
Maximum NRS baseline median 8 (range 9–1460) 24–2000)
1–10)‡
Engle et al36 – – – Infection within 29 d postoperation
n = 3 (2.7%) of 112 devices
implanted for cancer pain
Erdine et al37 – – – Urinary retention 21 (38.9)
Leakage-related headache 16 (29.6)
Nausea and vomiting 10 (18.5)
Pruritus 8 (14.8)
Infection 5 (9.3)
Disconnection 4 (7.4)
Migration 3 (5.6)
Neuromodulation 2023; 26: 1126–1141

Onofrio and Yaksh38 – Baseline ambulatory index: – –

Bedfast n = 21
Sits/walks sparingly n = 12
Walking n = 18
3 to 6 wk after implant:
Bedfast n = 14
Sits/walks sparingly n = 8
Walking n = 29
(Continued)
 www.neuromodulationjournal.org
Table 5. Continued

Author Pain intensity Functional, HRQoL outcomes and survival Systemic opioid intake Complications/side effects

Reig and Abejón 39

Baseline NRS median 8.23 (IQR 6–10)/FU 2.6 – – Test period complications
(IQR 1–10) in 22% of patients (most frequent post-
≥50% pain reduction = 78.12% dural puncture headache and urinary
retention); 16% mechanical complica-
tions (most frequent catheter dislodge-
ment); 8% surgical complications (mostly
postdural puncture headache); 33%
medication-related complications
(mostly nausea, vomiting, constipation)
Sayed et al40 Baseline (n = 152) median 7.1 (IQR 6–8)/1 mo Mean overall survival 138 d/median 65 d – Five patients (3.1%) had their pumps
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

(n = 83) 5 (IQR 2–6) explanted because of infection

International Neuromodulation Society. This is an open access article

↑ p < 0.0001
© 2022 The Authors. Published by Elsevier Inc. on behalf of the

Pain scores at 3-mo FU (n = 43; mean 4.47)

did not significantly differ from pain scores
at 1-mo FU (p = 0.384)/6 mo (n = 19) mean
4.11/12 mo (n = 7) mean 4.86
Scanlon et al41 – 51 patients died within 12-mo FU – Four patients (6.2%) had an SSI; of these,
three were pocket site infections and
one had a meningitis infection
Sindt et al42 – – Baseline mean ± SD daily MOED 305 ± 279 –
mg/FU 19 ± 57 mg
↑ p < 0.0001
94% reduction in daily MOED
Spiegel et al43 VAS baseline (n = 50) mean 6.6/1 mo (n = 38) 27 patients died within 6-mo FU Baseline median daily MOED 503 mg (range Numbness 8
4.4/3 mo (n = 29) 3.6/6 mo (n = 20) 3.4 35–5560)/1 mo 128 mg (0–3034)/3 mo 120 Urinary retention 7
mg (0–4320)/6 mo 105 mg (0–2880) Postdural puncture headache 5
Light-headedness 2
Sedation 1
Weakness 1
Yegul et al44 – – – Nausea and vomiting 10 (17.2)
Constipation 6 (10.3)
Headache 6 (10.3)

EFFECTIVENESS OF IDDS FOR CANCER PAIN

Urinary retention 5 (8.6)
Seroma 4 (6.9)
Pruritus 4 (6.9)
Infection 2 (3.4)
Neuromodulation 2023; 26: 1126–1141

Respiratory depression 2 (3.4)

(-) No statistically significant differences between assessments. ↑ Statistically significant improvements after IDDS.
ESAS, Edmonton Symptom Assessment System; FU, follow-up; IQR, interquartile range; SSI, surgical site infection; VRS, verbal rating scale.
*Individual participant data presented in Table 2 of Brogan and Winter29 for pain intensity and oral opioid intake. Individual participant data used for the purpose of meta-analysis.
‡
FU data from Dupoiron et al35 not presented separately for patients with implantable pump only except for overall survival.

1137
 DUARTE ET AL

Pain intensity month after treatment compared with baseline (MD, −3.53; 95%
For the primary outcome, nine reports, eight CI, −4.06 to −3.00; p < 0.001; I2 = 62.7%) (Fig. 2a). The mean
studies17,20,21,23,24,26,29,31 reporting data for 405 participants with an reduction in pain intensity was similar in prospectively designed
IDDS and one study33 reporting data for 42 participants with either an studies and retrospectively designed studies. The meta-analysis
IDDS or an SCS device, could be included in the meta-analysis of pain also including the study of participants with either an IDDS or an
intensity at the latest posttreatment follow-up time compared with SCS device (7 studies, 399 participants) showed similar results
baseline. The meta-analysis of the six studies including participants (Supplementary Data 5, Supplementary Data Fig. S6).
with an IDDS device shows a statistically significant reduction in pain Substantial heterogeneity (I2 between 62.7% and 96.5%) was
intensity (VAS or NRS) at the latest posttreatment follow-up time present between studies in the meta-analyses of pain intensity at
compared with baseline (MD, −3.31; 95% CI, −4.18 to −2.45; p < 0.001; up to one month and at the latest follow-up after implantation
I2 = 90.6%) (Fig. 2b). The meta-analysis also including the study of (Fig. 2a,b; Supplementary Data Figs. S1-S6). Heterogeneity did not
participants with either an IDDS or an SCS device (9 studies, 447 seem to be explained with subgroup analyses by study design
participants) showed similar results (Supplementary Data 5, (prospective or retrospective); results were similar within the two
Supplementary Data Fig. S5). subgroups for all analyses, and substantial heterogeneity remained
Seven reports, six studies17,20,21,24,29,31 reporting data for 325 within one or both subgroups for all analyses.
participants implanted with an IDDS and one study33 reporting The GRADE certainty of the evidence rating is low, with down-
data for 74 participants with either IDDS or an SCS device, could be grades to the certainty of the evidence made because of risks of
included in the meta-analysis of pain intensity up to one month bias within the studies included in the meta-analysis and because
after treatment compared with baseline. The meta-analysis of the of the substantial unexplained heterogeneity present. All studies
six studies including participants with an IDDS shows a statistically consistently demonstrate that pain intensity is significantly reduced
significant reduction in pain intensity (VAS or NRS) up to one following IDDS implantation compared with before implantation;
1138

Figure 2. Meta-analysis of pain intensity (a) at the latest follow-up following implantation of IDDS compared with baseline and (b) up to one month following
implantation of IDDS compared with baseline. REML, restricted maximum likelihood. [Color figure can be viewed at www.neuromodulationjournal.org]

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 EFFECTIVENESS OF IDDS FOR CANCER PAIN

however, the magnitude of pain reduction is uncertain because of was largely reduced, the authors concluded that there was insuf-
the high levels of unexplained heterogeneity present within the ficient evidence to establish the role of SCS in patients with cancer-
meta-analyses. related pain. SCS may be appropriate in cancer survivors with
residual neuropathic pain, but for patients with advanced cancer
with progressive disease, it is unlikely to provide pain relief for a
Other outcomes
new pain problem apart from the original pain, such as new bone
Overall, the outcomes reported in the included studies suggest
metastasis. In addition, SCS is known to be more effective for
that IDDS can lead to improvements in survival (compared with
neuropathic pain,45,46 whereas most types of cancer-related pain
CMM) and in HRQoL or functional outcomes (compared with before
will be mixed pain and may include a visceral element. Finally, SCS
implantation of IDDS). Cumulative survival in the RCT was 53.9% at
requires patient interaction for most devices through a patient-
six months compared with 37.2% for the CMM group.17 Other
handheld remote controller and charger; as such, it may not be
studies report survival using different methods, such as median
feasible for patients to operate an SCS toward end of life.
overall survival of 91 days22 or 142 days35; mean survival of 3.5
Only one RCT evaluating IDDS was identified,17 which was
months29; and postimplant survival of 39%, 24%, 16%, 11%, and 5%
judged to have a high RoB. The results of this RCT are quite his-
at 0.5, one, two, three, and ten years, respectively.26 Improvements
torical and may not reflect current oncologic practices or patient
in symptom severity and symptom interference with quality of
survival. The replication of an RCT several decades later could
life20,21 and HRQoL26 also were observed. One study investigated
address current evidence gaps. However, IDDS is a well-accepted
anxiety and depression, with no change in scores between baseline
management option for cancer pain, and as such, clinical equi-
and follow-ups.33
poise, essential for the conduct of an RCT, may not be present. RoB
All the included studies that assessed the use of systemic opioids
assessment of the nonrandomized studies also highlighted some
at baseline and follow-up showed a reduction in their use following
concerns. Most of the studies were conducted in a single center or
implantation of the IDDS. Three of the studies reported reductions of
single country, and therefore, the results may not be generalizable.
approximately 700 mg between baseline and follow-up.20,29,31 The
One large registry study (N = 1403) that may potentially reflect the
baseline morphine oral equivalent dose (MOED) in these studies
impact of IDDS in multiple centers and countries reported an
ranged from 796 mg to 1041 mg. A reduction in systemic opioids of
overall smaller impact on pain reduction than did single-center
286 mg was observed in another study, with a baseline MOED of 305
studies.26 However, most (81%) patients were from a single cen-
mg.42 One study reported that 92% of patients discontinued sys-
ter,26 limiting the generalizability of its findings.
temic opioids,21 whereas in another study, 70% of patients reported
All included studies that reported on the use of systemic opioids
a reduction of 50% or more from baseline.25 The mean doses
observed a reduction in its use following implantation of IDDS. This
reported at follow-up ranged from 5.35 mg24 to 307 mg.31
may well reflect heterogeneity of practice between centers, with
All but three studies35,38,42 reported complications or side effects
patients being referred for IDDS later in some centers than in
from IDDS or SCS. Complications/side effects are presented in
others, thereby resulting in higher morphine equivalent doses at
Tables 4 and 5. Postdural puncture headache was the most
baseline. However, even within the same center, there were
reported complication, whereas urinary retention, nausea, and
considerable differences in the baseline systemic opioid dose
vomiting were commonly reported side effects. One death was
reported across four separate studies in the years 2011 (796 mg/
reported following postoperative pneumonia,34 one death subse-
d),29 2015 (805 mg/d),20 and 2020 (305 mg/d,42 375 mg/d).21 It
quent to pulmonary embolism,22 and one death following pump
could be hypothesized that the lower systemic opioid dose in the
infection requiring explant and intravenous antibiotic therapy in a
latter reports could reflect changes in clinical practice and an
patient who subsequently died of acute renal failure from
attempt to avoid excessive use of opioids. However, another single-
obstructive uropathy.20 A study by Stearns et al26 reports that only
center study from 2020 reported mean systemic opioid doses at
two deaths were possibly associated with therapy; one was
baseline of 1041 mg/d.31
reported as infection with death secondary to postoperative
The proportion of patients discontinuing the use of systemic
pneumonia after device implantation, and one was because of
opioids or reductions observed in daily doses also may be affected
pulmonary embolus secondary to drug withdrawal as a result of
by different opioid reduction practices. Some of the US centers
missed pump refill.
convert all of the patient’s systemic medication into an intrathecal
opioid equivalent and abruptly discontinue all systemic opioids once
DISCUSSION the IDDS has been implanted.42 Such medically enforced discon-
tinuation of systemic opioids may limit the utility of systemic opioid
The results of our meta-analysis show statistically and clinically reduction as an outcome of IDDS therapy. Diverging practices
significant differences for reduction in pain intensity up to one toward systemic opioid reduction between US and European cen-
month after implantation of IDDS and at the latest posttreatment ters, where a more gradual, response-driven reduction in systemic
follow-up compared with baseline. Improvements in other out- opioids is common practice, may further limit the utility of this
comes following implantation of IDDS also were observed, whereas outcome.
serious adverse events because of IDDS do not seem to occur
frequently.
Only two studies included participants who received an SCS for Strengths and weaknesses
cancer pain.33,36 However, in these studies, the data also included This systematic review with meta-analysis evaluates the effect of
patients who received an IDDS, not allowing to ascertain the effects IDDS in the reduction of pain intensity for patients with cancer
of SCS for cancer pain. A Cochrane review evaluated the effec- pain. The meta-analysis results (ie, MDs and 95% CIs) and the pain
1139

tiveness of SCS for cancer-related pain compared with CMM and intensity reduction shown within most of the studies were
found four before-and-after case series.10 Although analgesic use consistently greater than the recommended minimal clinically

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 DUARTE ET AL

important difference of two points in the VAS or NRS.47 However, REFERENCES

substantial unexplained heterogeneity was present within the
meta-analyses, meaning that the magnitude of pain intensity 1. Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN esti-
reduction with IDDS therapy is uncertain. A greater heterogeneity mates of incidence and mortality worldwide for 36 cancers in 185 countries. CA
Cancer J Clin. 2021;71:209–249.
observed for reductions in systemic opioid intake prevented us 2. Bennett MI, Kaasa S, Barke A, et al. The IASP classification of chronic pain for ICD-
from carrying out a formal meta-analysis for this outcome. 11: chronic cancer-related pain. Pain. 2019;160:38–44.
The interpretation of the findings of this review and generaliz- 3. Hagedorn JM, Pittelkow TP, Hunt CL, D’Souza RS, Lamer TJ. Current perspectives
on spinal cord stimulation for the treatment of cancer pain. J Pain Res.
ability is limited by the availability of only one RCT and the 2020;13:3295–3305.
remaining nonrandomized studies providing mostly data from 4. van den Beuken-van Everdingen MH, de Rijke JM, Kessels AG, Schouten HC, van
Kleef M, Patijn J. Prevalence of pain in patients with cancer: a systematic review of
single centers. Limited evidence was identified evaluating the the past 40 years. Ann Oncol. 2007;18:1437–1449.
effectiveness of SCS for patients with cancer pain. The data were 5. Meuser T, Pietruck C, Radbruch L, Stute P, Lehmann KA, Grond S. Symptoms during
not reported in a consistent manner across the studies, preventing cancer pain treatment following WHO-guidelines: a longitudinal follow-up study of
symptom prevalence, severity and etiology. Pain. 2001;93:247–257.
the inclusion of further studies in the meta-analyses to evaluate 6. NHS England. Clinical commissioning policy: intrathecal pumps for treatment of
reductions in pain intensity and the feasibility of pooling data for severe cancer pain. Accessed August 17, 2021. https://www.england.nhs.uk/
other outcomes of interest. Because of limitations in study commissioning/wp-content/uploads/sites/12/2015/10/d08pb-intra-pumps-trtmnt.
pdf
reporting, a preplanned subgroup analysis could not be performed 7. Duarte RV, Sale A, Desai P, Marshall T, Eldabe S. The unmet need for intrathecal
to consider cancer-related pain or pain related with treatment for drug delivery pumps for the treatment of cancer pain in England: an assessment of
the hospital episode statistics database. Neuromodulation. 2020;23:1029–1033.
cancer. Furthermore, it was not possible to evaluate outcomes 8. Hayek SM, Deer TR, Pope JE, Panchal SJ, Patel VB. Intrathecal therapy for cancer
considering whether the patients had recovered from cancer or and non-cancer pain. Pain Physician. 2011;14:219–248.
had cancer progression. 9. Bruel BM, Burton AW. Intrathecal therapy for cancer-related pain. Pain Med.
2016;17:2404–2421.
10. Peng L, Min S, Zejun Z, Wei K, Bennett MI. Spinal cord stimulation for cancer-
related pain in adults. Cochrane Database Syst Rev. 2015;6:CD009389.
CONCLUSION 11. Xing F, Yong RJ, Kaye AD, Urman RD. Intrathecal drug delivery and spinal cord
stimulation for the treatment of cancer pain. Curr Pain Headache Rep. 2018;22:11.
Our results suggest that IDDS is an effective and safe manage- 12. Follett KA, Boortz-Marx RL, Drake JM, et al. Prevention and management of
intrathecal drug delivery and spinal cord stimulation system infections. Anesthe-
ment option and provides pain relief for patients with cancer pain. siology. 2004;100:1582–1594.
Statistically and clinically significant results were observed for 13. Centre for Reviews and Dissemination. Systematic reviews: CRD’s guidance for
reduction in pain intensity up to one month and the latest post- undertaking systematic reviews in health care. Accessed March 21, 2021. http://
www.york.ac.uk/inst/crd/SysRev/!SSL!/WebHelp/SysRev3.htm
treatment follow-up compared with baseline. The evidence also 14. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated
suggests that IDDS results in improvements in other outcomes. guideline for reporting systematic reviews. BMJ. 2021;372:n71.
However, the results should be interpreted with caution because of 15. Sterne JAC, Savović J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias
in randomised trials. BMJ. 2019;366:l4898.
the limitations of the evidence available. 16. National Heart Lung and Blood Institute (NHLBI). Study quality assessment tools;
2014. Accessed March 21, 2021. https://www.nhlbi.nih.gov/health-topics/study-
quality-assessment-tools
17. Smith TJ, Staats PS, Deer T, et al. Randomized clinical trial of an implantable drug
Authorship Statements delivery system compared with comprehensive medical management for refrac-
tory cancer pain: impact on pain, drug-related toxicity, and survival. J Clin Oncol.
Rui Duarte and Sam Eldabe conceptualized the study. Michelle 2002;20:4040–4049.
18. Smith TJ, Coyne PJ, Staats PS, et al. An implantable drug delivery system (IDDS) for
Maden conducted the searches. Rui Duarte, Sue Copley, and Ali refractory cancer pain provides sustained pain control, less drug-related toxicity,
Mohammed Al-Ali screened the search results for eligibility. Sue and possibly better survival compared with comprehensive medical management
Copley, Rui Duarte, and Sarah Nevitt extracted the data. Rui Duarte (CMM). Ann Oncol. 2005;16:825–833.
19. Smith TJ, Coyne PJ. Implantable drug delivery systems (IDDS) after failure of
and Sarah Nevitt conducted the risk of bias assessment. Sarah comprehensive medical management (CMM) can palliate symptoms in the most
Nevitt performed the data analysis. All authors contributed to drafts refractory cancer pain patients. J Palliat Med. 2005;8:736–742.
of the manuscript and approved the final version of the 20. Brogan SE, Winter NB, Okifuji A. Prospective observational study of patient-
controlled intrathecal analgesia: impact on cancer-associated symptoms, break-
manuscript. through pain control, and patient satisfaction. Reg Anesth Pain Med. 2015;40:
369–375.

 
21. Brogan SE, Sindt JE, Jackman CM, White J, Wilding V, Okifuji A. Prospective asso-
ciation of serum opioid levels and clinical outcomes in patients with cancer pain
How to Cite This Article treated With intrathecal opioid therapy. Anesth Analg. 2020;130:1035–1044.
22. Carvajal G, Dupoiron D, Seegers V, et al. Intrathecal drug delivery systems for
Duarte R., Copley S., Nevitt S., Maden M., Al-Ali A.M., refractory pancreatic cancer pain: observational follow-up study over an 11-year
period in a comprehensive cancer center. Anesth Analg. 2018;126:2038–2046.
Dupoiron D., Eldabe S. 2023. Effectiveness and Safety of 23. Cheng Y, Wang L, Zha Z, Zhao X. Effects of family integrated care on refractory
Intrathecal Drug Delivery Systems for the Management cancer-related pain receiving intrathecal morphine pump therapy: a cohort study.
of Cancer Pain: A Systematic Review and Meta-Analysis. Ann Palliat Med. 2020;9:2163–2170.
Neuromodulation 2023; 26: 1126–1141. 24. Dupoiron D, Lefebvre-kuntz D, Brenet O, et al. Refractory cancer pain and

 
intrathecal infusion: experience of three cancer management centers. Douleurs:
Evaluation - Diagnostic - Traitement. 2011;12:140–146.
25. Rauck RL, Cherry D, Boyer MF, Kosek P, Dunn J, Alo K. Long-term intrathecal opioid
therapy with a patient-activated, implanted delivery system for the treatment of
refractory cancer pain. J Pain. 2003;4:441–447.
SUPPLEMENTARY DATA 26. Stearns LM, Abd-Elsayed A, Perruchoud C, et al. Intrathecal drug delivery systems
for cancer pain: an analysis of a prospective, multicenter product surveillance
registry. Anesth Analg. 2020;130:289–297.
To access the supplementary material accompanying this article, 27. Kontopantelis E, Reeves D. Metaan: random-effects meta-analysis. The Stata
visit the online version of Neuromodulation: Technology at the Journal. 2010;10:395–407.
28. Langan D, Higgins JPT, Jackson D, et al. A comparison of heterogeneity variance
Neural Interface at www.neuromodulationjournal.org and at https://
1140

estimators in simulated random-effects meta-analyses. Res Synth Methods.

doi.org/10.1016/j.neurom.2022.03.003. 2019;10:83–98.

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 EFFECTIVENESS OF IDDS FOR CANCER PAIN

29. Brogan SE, Winter NB. Patient-controlled intrathecal analgesia for the manage- 44. Yegul I, Uyar M, Erhan E. Implanted drug delivery systems for spinal administration
ment of breakthrough cancer pain: a retrospective review and commentary. Pain of morphine in cancer pain relief. Agri-Istanbul. 1999;11:13–21.
Med. 2011;12:1758–1768. 45. Wolter T. Spinal cord stimulation for neuropathic pain: current perspectives. J Pain
30. Higgins JPT, Li T, Deeks JJ. Choosing effect measures and computing estimates of Res. 2014;7:651–663.
effect. In: Higgins JPT, Thomas J, Chandler J, et al, eds. Cochrane Handbook for 46. Deer TR, Mekhail N, Provenzano D, et al. The appropriate use of neurostimulation
Systematic Reviews of Interventions. 2nd ed. Cochrane; 2021. of the spinal cord and peripheral nervous system for the treatment of chronic pain
31. Chen GH, Spiegel MA, Magram YC, et al. Evaluation of fixed intrathecal bupivacaine and ischemic diseases: the Neuromodulation Appropriateness Consensus Com-
infusion doses in the oncologic population. Neuromodulation. 2020;23:984–990. mittee. Neuromodulation. 2014;17:515–550.
32. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating 47. Dworkin RH, Turk DC, Wyrwich KW, et al. Interpreting the clinical importance of
quality of evidence and strength of recommendations. BMJ. 2008;336:924–926. treatment outcomes in chronic pain clinical trials: IMMPACT recommendations.
33. Aigner CJ, Hernandez M, Koyyalagunta L, Novy D. The association of presurgery J Pain. 2008;9:105–121.
psychological symptoms with postsurgery pain among cancer patients receiving
implantable devices for pain management. Support Care Cancer. 2014;22:2323–2328.
34. Becker R, Jakob D, Uhle EI, Riegel T, Bertalanffy H. The significance of intrathecal
opioid therapy for the treatment of neuropathic cancer pain conditions. Stereotact COMMENTS
Funct Neurosurg. 2000;75:16–26.
35. Dupoiron D, Leblanc D, Demelliez-Merceron S, et al. Optimizing initial intrathecal
drug ratio for refractory cancer-related pain for early pain relief. A retrospective This review confirms the important role of neuromodulation tech-
monocentric study. Pain Med. 2019;20:2033–2042. niques in patients with cancer with pain poorly responsive to systemic
36. Engle MP, Vinh BP, Harun N, Koyyalagunta D. Infectious complications related to
intrathecal drug delivery system and spinal cord stimulator system implantations
treatment. Unfortunately, literature on intrathecal drug delivery is
at a comprehensive cancer pain center. Pain Physician. 2013;16:251–257. limited and the authors could find only one RCT; the literature on SCS
37. Erdine S, Oaeztakcin S, Yücel A. Intrathecal morphine delivered by implanted in patients with cancer is even poorer with indications only in iatro-
manual pump for cancer pain. Pain Digest. 1996;6:161–165.
38. Onofrio BM, Yaksh TL. Long-term pain relief produced by intrathecal morphine genic neuropathies. Nonetheless, the results of this review confirm the
infusion in 53 patients. J Neurosurg. 1990;72:200–209. good results achievable with intrathecal drug delivery.
39. Reig E, Abejón D. Continuous morphine infusion: a retrospective study of efficacy,
safety, and demographic variables. Neuromodulation. 2009;12:122–129.
40. Sayed D, Monroe F, Orr WN, et al. Retrospective analysis of intrathecal drug Laura Demartini, MD
delivery: outcomes, efficacy, and risk for cancer-related pain at a high volume Pavia, Italy
academic medical center. Neuromodulation. 2018;21:660–663.
41. Scanlon MM, Gazelka HM, Moeschler SM, et al. Surgical site infections in cancer ***
patients with intrathecal drug delivery devices. Pain Medicine. 2017;18:520–525. This article is important and confirms that IDDS has advantages over
42. Sindt JE, Odell DW, Dalley AP, Brogan SE. Initiation of intrathecal drug delivery other therapies, especially in high-dose opioid use.
dramatically reduces systemic opioid use in patients with advanced cancer. Neu-
romodulation. 2020;23:978–983.
43. Spiegel MA, Chen GH, Solla AC, Hingula LP, Legler A, Gulati A. Evaluation of an Jan Willem Kallewaard, MD, PhD
intrathecal drug delivery protocol leads to rapid reduction of systemic opioids in
the oncological population. J Palliat Med. 2021;24:418–422. Arnhem, The Netherlands

1141

www.neuromodulationjournal.org © 2022 The Authors. Published by Elsevier Inc. on behalf of the Neuromodulation 2023; 26: 1126–1141
International Neuromodulation Society. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).