Introduction
Due to local and systemic tumor effects and as a consequence of tumor-specific therapy,
30 % to 80 % of patients with tumors lose weight. Patients with tumors of the upper
gastrointestinal tract are particularly at risk of weight loss due to impaired chewing
and swallowing. Consequently, malnutrition correlates with impaired prognosis [1 ]. ESPEN Guidelines on Enteral Nutrition suggest that supportive nutrition therapy
should be started only if malnutrition is already present (BMI < 18.5) or if the patient
is unable to consume less than 60 % of the required daily energy intake over a period
of more than 7 to 10 days [2 ]. If low nutrient supply is expected to last longer than 2 to 3 weeks, placement
of a percutaneous endoscopic gastrostomy (PEG) should be favored over nasoenteric
systems [2 ]. One advantage of this method is the long viability and turnaround time largely
due to the safe fixation of the tube by intragastric and extragastric retention plates.
The most common PEG technique is the pull technique, which was first described in
1980 by Gauderer et al. After conventional esophagogastroduodenoscopy and percutanous
puncture of the stomach, a feeding tube with an intragastrically placed retention
plate passes the upper digestive tract [3 ]. As a modification, with the introducer technique, the tube is placed via Seldinger
technique using a guidewire.
A unique adverse event (AE) of PEG insertion in patients with head and neck cancer
is tumor seeding and formation of stromal metastases on the abdominal wall and viscera,
especially the stomach. That was first described by Preyer and Thul in 1989 [4 ]. In a review of all 44 reported cases, Cappell concluded that strong risk factors
for stomal metastases included primary pharynx-esophageal cancer, squamous cell histology,
less differentiated tumor, large size and advanced cancer stage, PEG placement by
the pull technique, untreated primary cancer with local recurrence and at a time > 3
months after PEG insertion [5 ]. Several authors have recommended use of direct endoscopic introducer technique
(Russell’s technique) or insertion under radiological guidance to help avoid this
dreaded AE [5 ]
[6 ]
[7 ].
Until now there have been only a few comparative studies related to PEG placement
by means of the pull technique or the direct puncture technique. Most of them were
retrospective and did not reveal consistent results [8 ]
[9 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ]. Almost all studies enrolled in a recent meta-analysis were observational studies
without randomization of insertion methods, and none of the groups were matched exactly
[15 ]. Although gastrostomy has been widely used, there is a lack of randomized trials
on this issue.
In accordance with the state of evidence at the time of initiation of our study, we
hypothesised that the puncture technique would decrease the frequency of inflammatory
AEs and rate of formation of metastases at the gastrostomy fistula in patients with
tumors in the upper gastrointestinal tract. Therefore, we investigated in an open
prospective trial whether application of the direct puncture technique (Freka-PEXACT-System)
is superior to the pull technique. Moreover, we aimed to compare the usage characteristics
in the long term.
Patients and methods
The primary endpoint was the rate of inflammatory AEs in the area of the gastrostomy
fistula. The working hypothesis postulated that use of the Freka-PEXACT-System during
placement of a PEG in patients with epithelial tumors of the upper gastrointestinal
tract or the larynx would reduce the rate of inflammatory AEs in the gastrostomy area
as compared to PEG placement using the pull technique as the PEG device would not
be in direct contact with the microbiome of the upper gastrointestinal tract. Under
the assumption that bypassing the upper gastrointestinal tract via the Freka-PEXACT-System
would reduce the frequency of metastases, the frequency of metastases in the stoma
area was defined as the secondary endpoint of the study.
The study was conducted as a monocenter, open, randomized, prospective trial at Leipzig
University. Recruitment took place over 18 months and started in January 2007. Inclusion
and
exclusion criteria are listed in Box . Assignment into the pull group (traditional PEG placement using the pull technique)
or into the puncture group (PEG placement using the Freka-PEXACT-System) was continuous
and changed on a monthly base. This resulted in well balanced patient assignments
to both groups ([Table 1 ]). The study participants were examined over a median period of 53 months (minimum/maximum
1 month/101 months, interquartile range 12.75; 96 months) after PEG placement. The
study protocol was voted on favorably by the Ethics Committee of the Medical Faculty
of the University of Leipzig.
Table 1
Patient characteristics showing no significant differences between groups.
Characteristics
Patients
Pull technique
Direct puncture
median (IQR)
median (IQR)
Age
71 (59 – 69)
67 (60 – 75)
BMI
23 (20 – 26)
23 (21 – 27)
n (%)
n (%)
Male
51 (86)
55 (84)
Indication
16 (27)
12 (18)
39 (66)
51 (78,5)
3 (5)
1 (1)
1 (2)
1 (2)
Location
16 (27)
23 (35)
31 (52)
31 (47)
10 (17)
10 (16)
2 (3)
1 (2)
BMI, body mass index
After patients’ approval via written consent for participation in the study and diagnostic
esophagogastroduodenoscopy, PEG placement was performed under sterile conditions and
2 g ceftriaxon were administered intravenously for antibiotic prophylaxis for 30 to
60 minutes during the pre-interventional period. A skin shave was done if appropriate
and the patients were placed in a supine position. We did not carry disinfect the
oral cavity or decontaminate the gut with antibiotics. Propofol sedation was used
in all cases. A 1-cm skin incision prior to insertion of the PEG was done after transillumination
in all patients. For the pull technique , we used Freka PEG 15 Charier (Fresenius Kabi AG, Bad Homburg, Germany) in accordance
with the manufacturer’s recommendations. Fixation of the PEG tube was reinforced by
means of an interior and exterior retention plate without sutures.
For PEG placement using the direct puncture technique , we used the Freka-PEXACT-System (Fresenius Kabi AG, Bad Homburg, Germany). In brief,
this required prior safe fixation of the stomach front wall to the abdominal wall.
For that purpose, 2 gastropexies were performed and placed laterally to the puncture
site using a gastropexy device in accordance with the manufacturer’s instructions.
Following the stab incision, a trocar and, subsequently, a pull-away sleeve were inserted
over which a balloon probe was placed and fixed by filling the balloon with 5 mL of
sterile water. Four weeks later, this was replaced with a permanent Freka Button 15
Charier (Fresenius Kabi AG, Bad Homburg, Germany, filled with 5 mL of sterile saline).
For that purpose, the gastropexy stitches were removed and the feeding tube was exchanged
percutaneously using a guidewire.
Follow-up for observation of early AEs was conducted on the Days 3, 7 and 21 post-intervention.
To include possible late AEs, further follow-up was carried out after 1, 3 and 6 months
and a last follow-up 5 years after intervention.
The occurrence of inflammatory AEs at the PEG placement site was determined both clinically
and anatomically. The occurrence of local signs of inflammation was also recorded.
Subjective patient satisfaction with the feeding tube was estimated using a visual
analogue scale (VAS) on which patients marked the level of their satisfaction along
a 100-mm straight line.
Computer-facilitated scoring and graphic representation of the collected data were
performed using Microsoft Office Excel 2007 and SPSS 15.0 for Microsoft Windows. The
following techniques were used for statistical evaluation: Mann-Whitney U test, chi-squared
test, Fisher’s exact test, Wilcoxon test, Kaplan-Meier survival analysis, logrank
test, as appropriate.
This trial was not registered through any of the registries approved by the International
Committee of Medical Journal Editors as this was not mandatory at the time of initiation
of our study in 2007.
Results
Patient characteristics
One hundred thirty-eight patients were screened for this study and 120 of them were
randomized ([Fig. 1 ]). The median age of the per protocol cohort was 64 years (IQR: 59; 74 years), 85 %
of study participants were men. The most frequent localization of the malignant disease
was the pharynx (50 %); followed by the oral cavity (31.5 %), the larynx (16.1 %)
and the esophagus (2.4 %). The median body mass index (BMI) was 23 (IQR: 21, 27).
The median initial Karnofsky Index was 80 % (40 – 90 %). There was no significant
difference in BMI and Karnofsky indices between the groups. 58 patients (48.3 %) were
assigned to the pull group, 62 patients (51.7 %) were assigned to the puncture group
([Table 1 ]).
Fig. 1 Patient randomization, allocation and per protocol intervention.
PEG procedure time
Procedure time for successful placement of the Freka-PEXACT-System was 20 minutes
(IQR: 15; 25 minutes) and for the pull technique was 15 minutes (IQR: 12; 20 minutes)
P < 0.001.
PEG placement using the pull technique was significantly faster in patients with BMI < 19
in comparison with the puncture technique (P = 0.09). No differences were found with regard to normal and overweight patients
([Fig. 2 ]).
Fig. 2 Box blot analysis of BMI vs. PEG procedure time. In underweight patients, the pull
technique was less time-consuming (P = 0.09).
The difference in median duration of PEG placement between the pull and the puncture
group was 10 minutes for stage T1 tumors (P = 0.002), 8 minutes for T2, 7 minutes for T3 and 3 minutes for T4 tumors ([Fig. 3 ]).
Fig. 3 Box blot analysis of tumor T-stage vs. PEG procedure time. The pull technique was
significantly faster only in T1 tumors (P = 0.002).
Tumor localization had no impact on procedure time (P = 0.87). The Karnofsky Index and a post-surgery state did not have any influence
on duration of PEG placement either. There was no statistically significant difference
in propofol dosages between the pull and the puncture technique (not shown).
PEG adverse events
Five patients suffered peri-interventional AEs. Two cases involved bleeding, which
was treated endoscopically. Three patients suffered a peri-interventional decrease
of oxygen saturation that necessitated increased oxygen supplementation, one of whom
required supplementary intermittent mask-administered respiration. No serious AEs
occurred.
Statistically significant differences emerged 1 month into the post-interventional
period, especially with regard to appearance of rubor (pull (n = 4) vs. puncture (n = 18),
P = 0.004). Pain at the placement site was found only in the puncture group (pull (n = 0)
vs. puncture (n = 5), P = 0.059). Other criteria of inflammation, such as swelling or the occurrence of pus,
did not differ significantly between both groups.
Overall, short-term AEs within 6 months occurred most frequently in the puncture group
(n = 33 vs. n = 14 in the pull group P = 0.0011 or P < 0.003). Specific AEs are shown in [Fig. 4 ].
Fig. 4 Percentage of short-term adverse events for the pull vs. the puncture PEG technique
within 6 months, *P < 0.005, otherwise not significant.
AEs did not emerge significantly more frequently in patients with cancer of the oral
cavity, especially with stage T4 tumors or when the tumor was not resected prior to
PEG placement. Patients who used the feeding tube did not experience AEs more frequently
than non-users.
Three months after PEG insertion and later, AE were rare. Between 3 and 6 months after
PEG insertion, 3 button PEGs (after Pexact removal) dislocated. The diameter of a
button PEG (filled with 5 cc sterile saline) was equal to the internal fixation disc
of the pull technique PEG (25 mm), as documented by the manufacturer. No relevant
PEG-related AEs were noted later than 6 months after PEG insertion. There were no
cases that showed any indication of metastasis formation at the gastrostomy fistula.
Patient satisfaction
Patient satisfaction with the percutaneous feeding tube was determined by means of
a visual analogue scale (VAS). Within the first 3 weeks, patients without AEs were
significantly more satisfied compared to patients with AEs according to their feedback
on the Days 3, 7 and 21 of follow-up (P = 0.033, P = 0.012, P = 0.040 respectively). We did not notice any significant differences between the
pull and the puncture group.
Long-term usage characteristics
After 3 months, 78 % of the study participants returned for medical examination; 6 %
died before their follow-ups could take place. After 6 months, the proportion of patients
examined was 60 %, whereas the percentage of deceased patients rose to 10 %.
The PEG tube was used without any essential differences between the study groups during
the 5-year follow-up period, with an overall usage, at least temporarily, by 101 of
the 120 patients (84.2 %). Time between PEG insertion and death was no different between
the pull and the puncture group (P = 0.63; [Fig. 5 ]). Total enteral nutrition via the PEG was applied in 84 patients (70 %). A scheduled
PEG removal after finishing multimodal tumor therapy was possible in 49 patients (40.8 %).
No significant differences between the pull and the puncture group were found with
regard to
frequency of scheduled PEG change (P = 0.95)
time between last PEG usage and death ([Fig. 5 ]) (P = 0.60)
PEG usage (P = 0.86)
total enteral nutrition via PEG (P = 0.079)
Fig. 5 The time between PEG insertion and death did not differ between the pull and the
puncture groups (P = 0.63).
Concerning increase in weight (BMI) and self-reliant care (Karnofsky Index), there
were no relevant differences between the pull and the puncture group during follow-up.
Discussion
In this prospective, open, randomized study, we report for the first time on long-term
PEG-related AEs and usage characteristics in a large cohort of patients with epithelial
tumors of the upper gastrointestinal tract. Contrary to our hypothesis, we demonstrated
that the innovative direct puncture device is associated with a higher rate of short-term
PEG-related AEs in comparison with the traditional pull technique. This was not associated,
however, with a reduced usage rate or impaired patient satisfaction.
Our prospectively observed patient cohort experienced a low rate of inflammatory AEs
(15.5 % and 21 % in the pull and the puncture group, respectively). This is lower
than reported in a recent retrospective study of 1625 patients (median follow-up 254
days) from 8 Korean hospitals, who received a pull technique PEG. In this study, 81.1 %
of all patients had a PEG inflammation [16 ]. This remarkable difference may be explained by the fact that our study used strict
prophylactic antibiotic supplementation [17 ] and the PEG procedure was carried out in a specialized endoscopy unit using established
standard operating procedures. We believe that our favorable results are not due to
underreporting as all patients were carefully followed by experienced nutrition assistants
and a standardized protocol was used.
The direct puncture technique was associated with a longer procedure time. This is
well in accord with a recently published retrospective comparative case series of
113 patients [18 ] and may simply be explained by the time-consuming performance of the 2 additional
gastropexies aside the PEG. Our detailed subgroup analysis, however, revealed that
the difference in procedure duration actually stems from the underweight and T1 tumor
groups. The pull technique was faster in low-weight patients with sparse visceral
fat and even small and less obstructing T1 tumors. Obstruction of the upper gastrointestinal
tract is a relevant factor only for the pull technique as all tumors had to be passable
by a standard gastroscope as per protocol. Consequently, the puncture technique was
equally time-consuming in all T stages. Longer intervention times in the puncture
group were not associated with higher propofol doses or peri-interventional adverse
events. Therefore, this difference will not be present in most patients with tumors
of the upper gastrointestinal tract and is of limited clinical relevance in low-weight
and T1 tumor patients.
A secondary endpoint of our study was the comparison of tumor seeding at the PEG insertion
site. Several authors have advocated the puncture technique on the grounds that it
prevented implantation metastases. However, case reports show that PEG insertion using
this technique does not eliminate the risk of direct tumor seeding [5 ]
[6 ]
[19 ]. A possible explanation for the occurrence of tumor seeding could be that the gastroscope
tip implants cells directly into the gastric wall during transillumination. It could
also be the result of swallowed tumor cells settling into the fresh gastric wound,
which would have caused the metastases to occur regardless of the method of PEG insertion.
Fortunately, none of the patients in our study developed a PEG metastasis. Nevertheless,
we support previous authors’ advice to avoid passage of the gastroscope deep into
the stomach. This may help to avoid direct contact with the gastric wall during transillumination
to prevent inadvertent seeding of the primary tumor to the PEG wound [19 ].
During the follow-up period of 4.5 years (median), randomization to the pull or the
puncture technique had no influence on the usage characteristics. About 5 of 6 patients
who received a PEG used it over the long term without any differences in usage characteristics
between the 2 devices used in this study. A scheduled PEG removal after finishing
multimodal tumor therapy was possible in more than 40 % of our patients. With regard
to quality of life, the puncture technique could have an advantage over the pull technique
as it does not require another endoscopy for PEG removal. However, that question has
not been addressed in this trial.
The Freka PEG, which was used for the pull technique, was manufactured with polyurethane,
whereas the Freka-PEXACT was manufactured with synthetic silicone rubber. Both materials
are highly bio- and tissue compatible. Theoretically, there is slightly more irritation
with silicone-based tubes because of the roughness of the surface, but in clinical
practice, insufficient sealing due to incorrect use or maintenance errors may be superimposed
on this factor. Differences between the materials, therefore, may be more reflective
of optical or functional aspects (color changes, occlusion etc.) than relevant to
AEs. A minor bias especially for the endpoint “rubor,” however, could be conceivable
[20 ]
[21 ].
We have not assessed glucose tolerance status in our patients. That may be a weakness
of our protocol as uncontrolled diabetes is a risk factor for wound infection. Albumin
was not measured systematically and there only BMI can be used as an objective measure
of nutrition. Howeve4r, BMI had no influence on any PEG complications.
Conclusion
In conclusion, our results favor the pull technique for application of a PEG in patients
with endoscopically passable tumors of the upper gastrointestinal tract due to fewer
short-term AEs. Both the pull and the puncture system contributed equally to secure
long-term use and patient satisfaction. The individual choice of an appropriate technique
should be based on the patient’s informed consent, individual factors such as BMI
or tumor size, and the operator’s experience and the facilities available.
Inclusion and exclusion criteria
Inclusion criteria
Evidence of a histologically confirmed malignant epithelial tumor of the upper gastrointestinal
tract or larynx
Tumor passable by standard gastroscope (external diameter ≥ 8.7 mm)
Evidence of a safe diaphanoscopy
Age ≥ 18 years
Written consent from patient or his/her legal representative for the placement of
a PEG and participation in this study after full procedural disclosure (24 hours before
the examination)
Exclusion criteria
Gastric varices
Coagulopathies
Peritonitis
Gastric ulcer in the front wall
Ascites
Status post-resection stomach surgery
Upper gastrointestinal tract stenosis including
patients with neurological dysphagia (alone or concomitant)
Pregnancy
