**4. Discussion**

To our knowledge, we are the first study to date to report our clinical findings of increased complications with the use of ALHMF in a Level IIIc clinical setting. In our retrospective analysis of acidosis, growth, and clinical outcomes in NICU infants fed with human milk fortified with LHMF and PHMF we found significant acidosis and poor growth in the infants receiving LHMF. These findings were very consistent with our clinical impressions during our clinical use of the LHMF. We were also surprised to see increased NEC in the ALHMF group. Although we were not powered as a primary outcome to evaluate NEC, we strongly encourage cautious further evaluation of the product in the clinical setting with regards to this serious outcome.

A key difference in the ALHMF as compared to the PHMF is the acidification process required for sterilization. This difference is likely to explain the increased complications seen in the ALHMF group. The preterm infant's inability to buffer this acid load likely led to an increase in clinical complications including acidosis, poor growth, and, possibly, NEC.

#### *4.1. Acidosis*

There was a higher incidence of clinically significant metabolic acidosis in the ALHMF group, with one infant requiring treatment with Bicitra. No infants in the PHMF group required Bicitra treatment, even with twice as many patients in this group. Premature infants are susceptible to metabolic acidosis [21] and renal tubular acidosis. However, these imbalances of acid base status should begin to normalize after the first weeks of extrauterine life [21]. Considering similar baseline characteristics, we hypothesize additional enteral acid load was a potential contributor to this increased incidence of metabolic acidosis in the LHMF group.

Premature infants are at risk for developing metabolic acidosis secondary to immature metabolic processes, a lower renal capacity to adequately excrete acid, and higher urinary losses of bicarbonate [2,4,22]. Quantity of protein may affect metabolic processes; however the median daily average protein provisions for each fortifier group were within the currently recommended ranges [1]. No clearly defined amount for maximum protein provision exists, however, it is suggested that intakes greater than 6 g/kg/day are poorly tolerated [2]. Maximum daily average protein provisions for both groups were below this level. Another reference states that protein provisions greater than 5 g/kg/day may cause azotemia [1], but each group had intakes below this value, and maximum BUN and creatinine levels were not different in the two fortifier groups. Having increased protein intake in the PHMF group as well as the ALHMF group helps to illuminate that increased protein content in the ALHMF was not likely the cause of the adverse outcomes.

We question if the acidification sterilization process of the ALHMF may contribute to this acidosis in some fragile premature infants. Our patient population included in this study was not limited by respiratory acuity, as was the population in the Moya *et al*. paper [23]. We hypothesize that our more inclusive population of both healthy infants and more fragile infants who may have less respiratory stability decreases their capability to buffer the acid load provided in the ALHMF resulting in clinical acidosis in some cases requiring medical therapy. It may be unwise in a fragile preterm infant population to minimize the clinical significance of the metabolic acidosis noted in the ALHMF groups in our study, the study conducted by Moya *et al*. [24], who reported that infants fed the ALHMF had significantly lower pH (at day six), bicarbonate (at day six and 14), and CO2 (at day 14 and 28), and significantly higher chloride (at day 14 and 28). Additionally, in an abstract evaluating 100 infants, 50 fed with ALHMF and 50 fed with PHMF, published by Cibulskis *et al*., from Saint Louis University at the 2013 AAP\_NCE, similar metabolic acidosis is described in this patient population (54% ALHMF *vs*. 10% PHMF, *p* = 0.0001) [25]. As reported in their abstract, this grouptreated the acidosis as if it were clinically significant, discontinuing ALHMF on 21/50 patients due to a clinical diagnosis of acidosis [25].

#### *4.2. Enteral Nutrition and Growth*

Infants in the PHMF group received a mean daily calorie intake of 117.3 kcal/kg/day as compared to infants in the ALHMF group who received a mean calorie intake of 128.7 kcal/kg/day. Infants in the ALHMF group also received a median of 0.4 g protein/kg/day more than the infants in the PHMF group. Despite higher protein and calorie provisions in the LHMF group, growth during the HMF period was slower between the two groups as evaluated by several methods: in a mixed effects model evaluated in g/kg/day (*p* = 0.002), in g/day (*p* = 0.0001), and by growth velocity in g/kg/day (*p* < 0.0001). Noted also, is that ALHMF infants experienced an additional decrease of 10 growth curve percentiles for weight from birth to 36 weeks EGA when compared to infants in the PHMF group (growth at 36 weeks is representative of nutrition delivery that is not limited to the period evaluated on PHMF and ALHMF). As Dexamethasone use inhibits growth in premature infants [4], we further note that fewer infants in the ALHMF group (5%) compared to the PHMF group (20%) required this drug for clinical treatment (*p* = 0.15).

Maintaining appropriate growth in this patient population was a high priority, so infants with suboptimal growth were fed increased caloric density feedings above 24 cal/oz. Though not statistically significant, a higher proportion of infants in the ALHMF group (48%) required caloric densities greater than 24 cal/ounce when compared to the PHMF group (26%). Had those 48% of infants in the ALHMF group not been prescribed increased caloric densities due to clinical observations of poor growth differences in growth throughout the hospitalization would likely have been larger between the PHMF groups and ALHMF groups. The statistical significance in infant growth as noted in g/kg/day is seen in spite of the high priority our unit takes in maintaining optimal growth and the subsequent aggressive adjustment of caloric density to achieve desired results. This was ultimately the reason 26% of infants receiving ALHMF were transitioned to receive the PHMF once the ALHMF use was discontinued in the NICU.

Not only are these growth effects consistent with the findings of Moya *et al*., they raise further questions [24]. Moya *et al*., reported no significant differences in rate of weight gain or head circumference growth between infants fed this same ALHMF and infants fed a PHMF, even though the human milk fortified with ALHMF contained 23% more protein (3.2 *vs*. 2.6 g protein/100 mL fortified preterm human milk) [24]. In our study, even though we compensated for the difference in protein content so that protein intake was similar, there was still poorer growth in infants fed ALHMF.

At least part of the inability of the additional protein to improve growth may be due to the acidosis noted above. It is well-known that infants with metabolic acidosis hyperventilate, as the expiration of CO2 drives the elimination of H+ ions through the bicarbonate buffering system. What is less well recognized is that protein catabolism can also be utilized to decrease acidity by the elimination of H+ ions through the urinary excretion of NH4 + . Acidosis reduces protein synthesis in rats [26] and leads to protein catabolism in humans [27].

#### *4.3. NEC*

No infants in PHMF group developed NEC compared to 13% in the ALHMF group. Reasons for these occurrences remain unclear, as similar prevention strategies were followed for each group. Previous implementation of aggressive nutrition practices in our unit demonstrated improved feeding tolerance and clinical outcomes, with no increased incidence of NEC [12]. These nutrition practices remained unchanged during the study period, and no additional clinical practices were implemented concurrent with the change in human milk fortification. Slow rate of enteral feeding advancement remained consistent between both groups, as evidenced by no statistically significant differences in length to full enteral feedings. No changes in brand or caloric density of premature infant formula were made, and infant formula was utilized equally in both groups if no MBM was available. As no additional practice changes were implemented during this study period, we can neither confirm nor exclude use of the ALHMF as a contributor to these occurrences of NEC. Although this study was not powered to detect NEC based on historical incidence in our unit with rates over the last five years ranging from 2% to 5% from our Vermont Oxford Network data, one should consider that significant differences with small sample sizes may either reflect coincidental effects due to sample size, or may be due to a real difference that is unexpectedly large.

#### *4.4. Metabolic Acidosis*

Literature suggests that premature infant formulas contain a high renal acid load, though human milk contains less [2,21]. Research has additionally demonstrated that the composition of infant formulas may affect the urinary pH and nutrient excretion of premature infants [21,22]. It is further proposed that high renal acid loads contribute to maximum renal acid stimulation (urine pH < 5.4) [28] in premature infants with immature renal function. Previous research studies have demonstrated that infants with metabolic acidosis or maximum renal acid stimulation exhibit decreased growth [28,29]. This may also result in an increase in urinary sodium excretion [24,29] and a decrease in nitrogen assimilation [30]. Blood sampling for acid-base indicators may not be significantly abnormal in the presence of maximum renal acid excretion [22,28]. However, CO2 values may trend low [28], which was clearly observed among infants in the ALHMF group (*p* = 0.002).
