Objectives

Our main objective was to identify the 30-day mortality rate and the procedural complications of gastrostomy insertion in patients with ALS in our center.

A secondary objective was to evaluate for the presence of any predictive factors of adverse outcomes, particularly regarding pulmonary function. Also, we wished to compare the post-procedural mortality and complications between percutaneous endoscopic gastrostomy (PEG) and radiologically inserted gastrostomy (RIG) procedures. Finally, we aimed to describe the survival and weight fluctuations in this population following gastrostomy.

Results

Mortality and complications

The 30-day mortality after the procedure was 9.3% (5/54 patients). These results were similar in both types of gastrostomy procedures. Of these five patients, three died from respiratory failure, and two transitioned to palliative care. In total, nine patients received medical assistance in dying. However, none of these occurred in the 30-day period following the gastrostomy procedure.

Table 4 details the 30-day mortality rate and the incidence of the different complication categories for both PEG and RIG procedures.

Table 4. Procedural complications and mortality

Major complications following the procedure occurred in nine patients (16.7%). Of these, three patients (5.6%) had respiratory failure, two patients (3.7%) had aspiration pneumonia, one patient (1.9%) had gastric wall ischemia, one patient had tube leakage causing pneumoperitoneum, one patient had gastrointestinal hemorrhage and one patient had an abscess. In total, 28 patients (51.9%) suffered from at least one complication. The different complications that occurred in this study are detailed in Appendix 1.

Predictive factors for adverse outcomes

The logistic regression analysis performed to identify any predictive factors of complications is detailed in Appendix 2. Table 5 summarizes only the results of the logistic regression, which seem statistically significant. For all complications, pulmonary disease had an odds ratio of 4.26 (1.11–21.14), and hypercapnia had an odds ratio of 1.09 (1.01–1.21). For major complications, diabetes had an odds ratio of 5.13 (0.84–29.64), thrombocytosis (elevated platelet levels) an odds ratio of 1.01 (1.00–1.02) and hypercapnia an odds ratio of 1.13 (1.03–1.28). For respiratory complications, diabetes had an odds ratio of 5.13 (0.84–29.64), and hypercapnia had an odds ratio of 1.15 (1.04–1.31). For gastrointestinal complications, pulmonary disease had an odds ratio of 4.81 (1.28–19.21), and the presence of a respiratory therapist had an odds ratio of 6.00 (1.26–33.63). Finally, for mechanical complications, RIG had an odds ratio of 6.33 (1.40–30.28), and the presence of a respiratory therapist had an odds ratio of 5.57 (1.10–29.28).

Table 5. Strength of association* with risk factor according to logistic regression analysis for complications following gastrostomy

Note: RIG = radiologically inserted gastrostomy. *Results close to statistical significance (p < 0.05) are also included in the table.

As described previously, logistic regression analysis could not be performed for mortality, infectious complications and hemorrhagic complications, as the numbers were too low. Thus, Fisher’s exact test and Wilcoxon rank sum test were conducted to investigate specific predictive factors of such complications, but none yielded statistically significant results (see Appendix 3).

Discussion

In our study, the 30-day mortality rate following gastrostomy is 9.3%, which is in the wide range of 2%–25% described in the literature for ALS patients. ^{Reference Castanheira, Swash and De Carvalho8,Reference Sulistyo, Abrahao, Freitas, Ritsma and Zinman16–18} As expected, the majority of these deaths (3/5 patients) are from respiratory failure, with the others being from transition to palliative care. Mortality rates are similar between percutaneous endoscopic and radiologically inserted gastrostomy (9.1% vs 10%).

The incidence of major complications is 16.7%, which is also similar to the incidence of 2%–16% described in the literature. ^{Reference Sulistyo, Abrahao, Freitas, Ritsma and Zinman16} This major complication rate is in the upper range of the comparative to the literature. In our opinion, this is explained mostly by the fact that we were very inclusive in our definition of complications being labeled as “major.” Additionally, being a small volume center (54 patients in 13 years) can lead to a higher rate of complications. The most common complications in our study are benign pneumoperitoneum, buried bumper syndrome, tube leakage, respiratory failure, aspiration with or without pneumonia and peristomal cellulitis. In total, 51.9% of patients suffered from at least one complication. These data seem consistent with the literature, although precise values in this particular population seem to vary widely from one reference to the other. For example, one reference suggests that the rate of complications from the “pull” method is 74%, while the rate for the “introducer” method is 47%. ^{Reference Sidaner17} In our center, the former is used for PEG insertion and the latter for the RIG procedure.

In the available literature, there is certain evidence suggesting that RIG is more advantageous than PEG since it requires less sedation and could thus be safer in patients with lower FVC values. ^{Reference Castanheira, Swash and De Carvalho8,Reference Russ, Phillips, Wilcox and Peter32–Reference Stavroulakis, Walsh, Shaw and McDermott34} This finding is inconsistent as other studies suggest that both procedures are equivalent in terms of safety, regardless of pulmonary function. ^{Reference Desport, Mabrouk, Bouillet, Perna, Preux and Couratier13,18,Reference Allen, Chen and Ajroud-Driss35–Reference Yuan, He, Wang, Kong and Cao37} In our study, complications incidence appears to be slightly higher in the RIG group than the PEG group in terms of major complications (20% vs 15.9%), total complications (70% vs 47.7%), respiratory complications (20% vs 15.9%), hemorrhagic complications (20% vs 0%) and mechanical complications (50% vs 13.6%). This can be explained by the fact that in our center, it is usually the technically more challenging cases or generally sicker patients (e.g., patients with more advanced disease) who are referred to have their gastrostomy in radiology.

In previous studies, certain factors have been associated with a higher risk of complications or mortality following gastrostomy insertion in the ALS population. These include a high C-reactive protein level, ^{Reference Blomberg, Lagergren, Martin, Mattsson and Lagergren38} low albumin, ^{Reference Sidaner17,Reference Blomberg, Lagergren, Martin, Mattsson and Lagergren38,Reference Nunes, Santos, Grunho and Fonseca39} low transferrin, ^{Reference Nunes, Santos, Grunho and Fonseca39} low body mass index, ^{Reference Castanheira, Swash and De Carvalho8,Reference Nunes, Santos, Grunho and Fonseca39} weight loss before the intervention, ^{Reference Chiò, Finocchiaro, Meineri, Bottacchi and Schiffer5,Reference Castanheira, Swash and De Carvalho8,Reference Sidaner17} advanced age ^{Reference Chiò, Finocchiaro, Meineri, Bottacchi and Schiffer5,Reference Castanheira, Swash and De Carvalho8,Reference Sidaner17} and elevated CO₂ levels. ^{Reference Bokuda, Shimizu and Imamura20} The FVC has also been identified in some studies as a factor related to complications, particularly respiratory failure and pulmonary aspiration. ^{Reference Chiò, Finocchiaro, Meineri, Bottacchi and Schiffer5,Reference Kasarskis, Scarlata, Hill, Fuller, Stambler and Cedarbaum19,Reference Bokuda, Shimizu and Imamura20,Reference Dorst, Dupuis and Petri23} For example, Pena et al. ^{Reference Pena, Ravasco and Machado40} have shown that FVC < 50% is associated with greater mortality following PEG. However, as Castenheira et al. ^{Reference Castanheira, Swash and De Carvalho8} have shown, this finding is inconsistent from one study to another, with some data suggesting that the procedure is safe in patients with an FVC lower than 50% and even in those with an FVC of less than 30%.

In our study, the following factors are statistically and clinically significant: pulmonary disease and total complications, hypercapnia and major complications and hypercapnia and respiratory complications. Thus, pulmonary disease and hypercapnia seem to be important predictive factors of complications following gastrostomy in patients with ALS. Therefore, these elements should be taken into consideration before the procedure.

SVC was not a significant factor associated with a heightened risk of complications in our study. This suggests that the traditional 50% vital capacity level could be too conservative and should be re-examined in future studies (i.e., there likely is a ventilatory risk level, but perhaps lower than 50%). However, it should be noted that only 38/54 patients (70.4%) had an SVC measured, which could represent a significant bias in our study. It is possible that these patients with missing data would have had lower SVC values (rapidly progressing ALS or more advanced bulbar involvement limiting pulmonary function tests). It is possible that only an arterial blood gas was performed for these patients. Thus, they were not included in the logistic regression for SVC but could have been included in the logistic regression for CO₂ levels, which could explain why the latter was identified as a predictive factor and not the former. This bias is an important one to keep in mind when interpreting the results of this study.

Furthermore, the following factors are clinically significant and are close to statistical significance: hypercapnia and total complications, diabetes and major complications and diabetes and respiratory complications. As described above, hypercapnia does seem to be a predictive factor for complications. In addition, although not statistically significant, there is a signal for diabetes as a predictive factor for complications.

The following factors were of statistical significance, but of uncertain clinical significance: platelet count and major complications, pulmonary disease and gastrointestinal complications, presence of a respiratory therapist during the procedure and gastrointestinal complications, RIG and mechanical complications and presence of a respiratory therapist and mechanical complications. The association for platelet count was minimal (OR = 1.01), which undermines its relevance. Although the clinical relation between pulmonary disease and gastrointestinal complications is unclear, pulmonary disease was shown to be a factor associated with all complications following gastrostomy in our study population. Regarding the presence of a respiratory therapist for the procedure, this could be understood as a surrogate for the patient having a more severe disease or comorbidities that heighten the risks of the procedure. Finally, as stated earlier, it is the more technically challenging cases that are referred to radiology in our center. This could explain the association between this type of gastrostomy and the occurrence of mechanical complications.

Finally, median survival (where T0 = time of diagnosis) was calculated to be 1.77 years when including patients having received medical assistance in dying and 1.91 years when censoring such patients. This is similar to what has been described in the literature, where life expectancy with ALS is said to be between 20 and 48 months. ^{Reference Chio, Logroscino and Hardiman2} We have chosen to demonstrate survival both including and censoring patients having received medical assistance in dying because it represents real-world data.

Finally, our study has several limitations to consider. First of all, the retrospective nature of this study is a limitation inherent to its design, compared to a prospective trial. Consequently, certain patients were eventually lost to follow-up, so their outcomes are unknown. Although such patients were censored from survival analysis, missed adverse events could have an impact on the results, notably underestimating mortality rates and incidence of complications, as well as affecting the reliability of the logistic regression on predictive factors of such complications. This is particularly important with regard to the SVC, where missing information could represent more severe ALS, such as advanced bulbar or pulmonary disease. Another bias concerning the SVC is the time between the pulmonary function test and the gastrostomy procedure, which could be unreliable if there is a significant delay between both. As previously mentioned, we chose a cutoff of 6 months before the procedure or 3 months after the procedure for the pulmonary function test to be considered valid and included in the statistical analysis. This is a large time-window and could represent a source of bias as the actual SVC value at the time of the procedure could actually be different from the one measured at the time that the pulmonary function test was initially performed. The reason for this choice of a long time-window was to include as many patients as possible in this analysis (since already only 70.4% of patients had an SVC measure available), although we understand that it introduces another significant source of bias. Another bias that impacts the power of our study is the small sample size of our population (n = 54). Finally, the logistic regression performed on a wide variety of elements could be considered a form of “data dredging.” However, this was included in the protocol from the beginning of the study and is necessary to achieve our objectives. Moreover, positive results are interpreted critically to confirm their clinical relevance. Furthermore, we believe that our results are comparable to the literature, so they can be generalized to the bigger population.

Conclusion

In our study population, the 30-day mortality rate following gastrostomy in patients with ALS was 9.3% (5/54 patients), and the incidence of major complications following the procedure was 16.7% (9/54 patients).

Identified predictive factors of complications following gastrostomy that are statistically significant and appear to be most clinically significant are pulmonary disease and blood CO₂ levels. There was a signal, although not statistically significant, for diabetes as a predictive factor. The SVC was not a predictive factor of complications in our study population, although only 70% of patients had a measure of SVC, which introduces a significant bias.

The results of our study provide additional evidence that the traditional 50% vital capacity level could be too conservative and should be re-examined (there likely is a ventilatory risk level, but perhaps lower than 50%). Further studies should be pursued in order to verify this suggestion and identify a clearer cutoff. Furthermore, precautions should certainly be taken in patients with reduced pulmonary function to limit the occurrence of complications such as minimizing sedation, providing noninvasive ventilation as required during the procedure and having a respiratory therapist present.