The primary aim of this study was to assess the changes in total serum AP activity around cardiothoracic surgery in infants. Our study is the largest to date to examine serum AP activity in any patient population undergoing cardiothoracic surgery. In 1974, Neutze et al
. briefly described a post-operative decrease in total serum AP activity in 15 children undergoing cardiac surgery with profound hypothermia and limited CPB [19
]. The authors did not further explore the etiology or clinical changes associated with decreased AP. Lum et al
. followed activity in 25 adult CPB patients and showed a 48% decrease in AP activity relative to baseline, compared to a decrease of approximately 15% in non-cardiac surgery patients [18
]. Serial AP activity demonstrated a nadir on POD1 with subsequent recovery over the next 7 to 10 days.
In our population, pre-operative AP activity varied significantly between surgical groups. Specifically, the CPB group showed higher pre-operative activity than the non-CPB and DSC groups. These differences were anticipated given the older age of the CPB cohort and the normal age variations in AP activity during infancy [3
]. However, it is possible that additional factors such as pre-operative cardiovascular instability and inflammation could modulate pre-operative AP activity. While no patients received enteral feeding within six hours of surgery, enteral feeding status in the days prior to surgery may also have played a role.
Despite these pre-operative differences, we found that AP activity decreased following surgery in almost every case. Often this drop was quite profound. DSC patients showed the greatest relative drop in AP activity with an average decrease > 60%. We also found that these changes were not exclusive to infants exposed to CPB. While the decrease was generally smaller, eight out of nine patients in the non-CPB group exhibited a decrease in their AP activity. The one patient who underwent DSC without CPB exhibited a greater AP decrease of 132 U/L, suggesting that large decreases in AP activity following surgery cannot be purely ascribed to the use of CPB.
Our study was not designed to determine the etiology of decline in AP activity following cardiothoracic surgery. However, we are able to make three preliminary observations. First, removal of AP by the circuit during hemofiltration is an unlikely etiology, as AP is a large molecule (> 65,000 Daltons) and the Sieving coefficient through a polyethersulfone membrane is effectively zero. Second, in contrast to the findings of Lum et al
], total protein and albumin increased in our patients following surgery, and changes in total protein and albumin did not correlate with changes in AP. These findings make dilution less likely as an etiology, although albumin replacement was not monitored in this study. Third, liver injury is also unlikely as a mechanism, as AP activity was inversely related to ALT, the opposite of the typical pattern in liver congestion or injury [1
The secondary objective of our study was to examine the relationship between decreased AP activity to requirements and duration of post-operative support, and short term clinical outcomes. In our population, lower total AP activity on POD1 was independently associated with increased intensity and duration of post-operative support, including higher vasoactive and inotropic medication requirements, prolonged intubation time, and increased length of hospital stay. The change in AP activity was not as significant as the absolute activity on POD1, in predicting the need for post-operative support. While the predicted change in support for each unit decrease in AP activity is relatively small, the large differences in post-operative AP activity make these predicted changes clinically relevant. For example, comparing patients at the 25th and 75th percentile in the CPB group (AP 77 vs. 166 U/L), those in the 25th percentile would be expected to have an approximate increase in VIS of 6 (equivalent to the addition of 6 mcg/kg/minute of dopamine or 0.06 mcg/kg/minute of epinephrine), a 50% increase in intubation time, and a 40% increase in length of hospital stay. We did note a small trend towards increased odds of cardiac arrest or death in patients with lower AP activity. In particular, AP activity below 90 U/L appeared to be a sensitive predictor of cardiac arrest or death in our population, and might be a candidate threshold for future observational and interventional research. However, the study size was insufficient to draw any definitive conclusions regarding post-operative AP activity and cardiac arrest or death.
Given these associations with post-operative support and the trend towards differences in the frequency of poor outcomes, AP activity may have some utility as a predictor of increased post-operative support. We did not, however, examine its predictive strength compared to other available biomarkers (such as lactate trends) or scoring systems (such as VIS) already proven to predict differences in outcome. Of greater potential interest, however, is the question of whether AP activity is simply a marker of disease severity, or if AP has a true biologic role in modulating the response to cardiothoracic surgery. Due to the observational design of the study, we cannot directly draw conclusions about cause and effect. Recent literature, however, has suggested that AP may be active in multiple inflammatory and ischemic settings [2
]. AP displays various potentially beneficial actions including modulation of nitric oxide synthase [13
] and dephosphorylation of extracellular ATP [23
]. The link between AP and inflammation is also a topic of active research. AP is capable of dephosphorylating the lipid-A moiety of endotoxin, converting it to a non-toxic monophosphoryl product [2
]. AP may also target additional bacterial components such as CpG DNA and flagellin [4
]. In cardiac surgery, endotoxin release from the intestine is thought to be one trigger for systemic inflammation, and AP may play a role in reducing this inflammation [8
]. To assess if there was a potential link between AP and systemic inflammation in our population, we examined the association between AP and the two inflammatory biomarkers available for analysis through the parent PCT study (PCT and CRP). We were able to demonstrate that lower AP activity on POD1 was strongly associated with increased PCT but not CRP levels. PCT is highly activated by endotoxin [24
], and prior pediatric studies of cardiac arrest and CPB have shown that elevated levels of PCT, but not CRP, are predictive of poor outcome [24
]. In this study AP activity was also inversely correlated with PCT levels before the operation, generally in the setting of poor systemic blood flow and lactic acidosis. These data provide some initial evidence for a possible biologic mechanism for AP in our population.
If low AP activity decreases the protective host response to cardiothoracic surgery, then supplementation of AP to higher levels could potentially allow for a reduction in post-operative support requirements. Evidence in favor of AP as a therapeutic agent in various disease processes has recently been established. Enteral administration of AP has been shown to improve outcomes in rat models of necrotizing enterocolitis and Crohn's Disease [10
]. Administration of Resolvin-E1 in a rat model of Crohn's disease increased expression of intestinal AP and reduced disease activity, improvements which were blunted with inhibition of intestinal AP activity [5
]. Administration of exogenous AP has also been effective in animal models of systemic inflammation [6
Several human trials have been performed using AP supplementation. An uncontrolled trial in adults with ulcerative colitis showed that administration of oral AP for seven days resulted in decreased CRP levels and improved clinical response scores [7
]. Heemskerk, et al
. assessed the effects of AP on renal injury in adult patients with severe sepsis/septic shock [13
]. AP-treated patients showed an improvement in serum creatinine and glutathione S
-transferase A1-1 compared to controls, possibly due to a demonstrated reduction in the expression of renal nitric oxide synthase. In a follow-up study, Pickkers, et al
. showed improvement in a combined renal function outcome as well as a reduction in systemic inflammatory markers in the AP group compared to placebo [23
]. In addition to the effects of AP on endotoxin, this group also suggested a role for AP in the dephosphorylation of extracellular ATP. Kats, et al
. evaluated the effects of exogenous AP on inflammation and outcomes in adult patients undergoing coronary artery bypass grafting [8
]. Given the overall favorable clinical course, the study was underpowered to assess a difference in surgical outcomes. However, AP-treated patients did not demonstrate the post-operative rise in the inflammatory markers TNF α, IL-6, and IL-8 seen in a portion of control patients. The authors hypothesized that AP might play a role in preventing post-bypass inflammation, potentially by limiting endotoxin leak from the hypoperfused gastrointestinal tract.
In summary, our study confirms that the decrease in AP activity seen in prior studies reliably occurs in infants undergoing cardiothoracic surgery. Low AP activity was, in turn, associated with the need for increased post-operative support. Patients with low post-operative AP activity may also have an increased risk of cardiac arrest or death, but our study was underpowered to confirm this association. Low post-operative AP activity was significantly associated with elevated PCT levels, providing limited initial evidence for a link between AP and one pathway of systemic inflammation. Based on these results, we believe serum AP may represent an important mediator and potential therapeutic target following cardiothoracic surgery in infants.
There are several limitations to our study. First, it represents a single center experience and generalizability to other centers has not been established. Second, the protocol was designed to answer questions about procalcitonin and CRP rather than AP. The research was therefore restricted by the structure and sample size of the parent study. The sample size was underpowered to assess more rare events such as mortality, cardiac arrest, and the need for mechanical support. Small sample size also limited the scope of covariate assessment in our multivariate analysis. The sub-analysis design limited the collection of additional data of interest, such as specific changes in isoforms, the timing of AP recovery, broader inflammatory biomarker analysis, and endotoxin assays. These aspects should be addressed in future studies. One specific issue of the parent study design concerns the protocols for assessing AP activity. No AP activity was sampled in the very early post-operative period (0 to 6 hours), and data could have been missed on the kinetics, etiology, and relationship to early (POD 0) deaths. POD 1 AP activity was not mandated in the non-CPB group resulting in three patients who could not be analyzed, and potentially limiting the conclusions that can be drawn from this group. Also, all AP activity was assessed at the same time point on POD1 regardless of the time of surgery, potentially introducing an element of bias if surgery performed later in the day differed substantially from morning surgery. Finally, as mentioned previously, this study is observational in design and cannot establish causality. Further research is required to assess whether modulation of AP activity in the peri-operative period will affect post-operative support, inflammation, or outcomes.