Manifestations of Cow's-Milk Protein Intolerance in Preterm Infants : Journal of Pediatric Gastroenterology and Nutrition

Secondary Logo

Journal Logo

Original Articles: Nutrition

Manifestations of Cow's-Milk Protein Intolerance in Preterm Infants

Cordova, Jonathan; Sriram, Sudhir; Patton, Tiffany; Jericho, Hillary; Gokhale, Ranjana; Weinstein, Dana; Sentongo, Timothy

Author Information
Journal of Pediatric Gastroenterology and Nutrition 62(1):p 140-144, January 2016. | DOI: 10.1097/MPG.0000000000000933
  • Free



Cow's-milk protein intolerance (CMPI) is poorly recognized in preterm infants. This study examined the clinical events that preceded the diagnosis of CMPI in preterm infants.


This was a retrospective study of infants in a level-III neonatal intensive care unit of those who received parenteral nutrition (PN) support during a 12-month period. Parameters assessed included birth weight (g), diagnosis, duration and frequency on PN, type of enteral feeds at initiation, and achievement of enteral autonomy. CMPI was diagnosed based on persistent feeding intolerance that resolved after change of feeds from intact protein to a protein hydrolysate or crystalline amino acid formula.


Three hundred forty-eight infants with birth weight (median/range) 1618 g (425–5110) received PN. Fifty-one (14%) infants required multiple courses of PN, and 19 of 348 (5%) were diagnosed with CMPI. The requirement for multiple courses on PN versus single course was associated with a high likelihood of CMPI: 14 of 51 versus 5 of 297, P < 0.001. Nine of the 14 infants identified with CMPI were initially diagnosed with necrotizing enterocolitis (NEC) after a median duration of 22 days (19–57) on intact protein feeds. After recovery from NEC, they had persistent feeding intolerance including recurrence of “NEC-like illness” (N = 3) that resolved after change of feeds to a protein hydrolysate or crystalline amino acid formula.


The requirement for multiple courses of PN because of persistent feeding intolerance after recovery from NEC and recurrence of “NEC-like illness” may be a manifestation of CMPI in preterm infants.

What Is Known

  • Cow's-milk protein intolerance affects 3% to 7% of healthy children.
  • Cow's-milk protein intolerance is poorly described in preterm infants.
  • Necrotizing enterocolitis is the most common gastrointestinal emergency in preterm infants and classically presents during the second week of life following the introduction of feeds.

What Is New

  • Manifestations of cow's-milk protein intolerance in preterm infants include recurring feeding intolerance and the requirement for multiple courses of parenteral nutrition therapy.
  • Cow's-milk protein intolerance was associated with prolonged hospitalization in preterm infants.
  • Changing feeds to a protein-hydrolyzed or crystalline amino acid–based formula was effective therapy in preterm infants with recurring feeding intolerance after necrotizing enterocolitis.

Parenteral nutrition (PN) is now established as the routine initial source of nutrients for preterm, critically ill newborn infants, and those with severe congenital or acquired illness that impairs function or the use of the gastrointestinal tract. The high incidence of respiratory problems, limited gastric capacity, and intestinal hypomotility in small preterm infants dictates the need for slow advancement of the volume of enteral feeds (1). The goal of PN support is to enable nutrient intake at rates that support positive nitrogen balance at normal postconception rates of growth while the infant is adapting to enteral feeds.

Cow's-milk protein intolerance (CMPI) is an inflammatory gastrointestinal process that affects 3% to 7% of healthy infants during the first year of life and typically presents following a period of exposure and sensitization to Cow's-milk protein–based infant feeds (2). The clinical manifestation is protean with symptoms ranging from persistent colic to reflux, feeding difficulty, diarrhea, mucus stools, hematochezia, ileus, and intestinal dysmotility resembling Hirschsprung disease (2,3). A positive diagnosis is confirmed by the resolution of symptoms after the elimination of dairy consumption by nursing mothers or change of infants’ feeds to a protein hydrolysate (PH), such as Pregestimil (Mead Johnson Nutrition, Glenview, IL) or a crystalline amino acid (CAA)–based formula, such as Elecare (Abbott Nutrition, Columbus, OH) or Neocate (Nutricia, Gaithersburg, MD). The incidence and clinical manifestations of CMPI in preterm infants with comorbidities remains poorly defined. Necrotizing enterocolitis (NEC) is the most common gastrointestinal emergency in preterm infants, affecting ∼7% to 11% of very-low-birth-weight infants, and the incidence is not expected to decline (4–6). The “classic” presentation is the onset of abdominal distension, increased gastric residuals, hematochezia, and clinical instability subsequent to the initiation of enteral feeds within the first 2 weeks of life (5,7). The pathognomonic findings can include pneumatosis intestinalis and/or portal venous gas on abdominal radiographic x-ray (5). Almost 50% of affected infants who develop NEC will require emergent surgical intervention as a result of bowel necrosis and perforation (8).

During routine monitoring of preterm infants on PN therapy, we observed a recurring pattern of extended duration on therapy or the requirement for multiple courses of PN because of persistent feeding intolerance, late-onset NEC, and repeat NEC-like episodes that resolved after the change of feeds to a PH- or CAA-based formula. Hence, from a nutrition support perspective, persistent feeding intolerance, late-onset NEC, and recurring NEC-like episodes need to be investigated because of their frequent need for management with prolonged duration on PN and/or multiple courses of PN therapy. Therefore, we sought to explore the association between this clinical presentation and subsequent diagnosis of CMPI in infants.


This was a retrospective study of all newborn and preterm infants admitted to a level III neonatal intensive care unit who received PN therapy during a 12-month period. All the newborn infants judged medically unstable for safe administration of full enteral feeds were started on PN support within 12- to 48-hours of admission to the neonatal intensive care unit. After achieving medical and/or postsurgery stability, enteral feeding with human milk, or gestational age appropriate standard intact protein infant formula was initiated at trophic rates. Feeds were administered via nasogastric tube in those infants unable to orally feed because of prematurity, sedation, or illness-related impediments in oral feeding function. The enteral feeds were then gradually advanced as tolerated with corresponding adjustments of PN volume and composition to ensure adequate protein, energy, and fluid intake.

The duration and number of courses of PN therapy was assessed. A single course of PN was defined as daily receipt of PN with interruptions lasting <5 days. The reinstatement of PN therapy after an interruption of >5 days was considered a new course of PN. The clinical parameters assessed included gestational age (weeks), birth weight (grams), clinical diagnosis, and growth in weight, length (centimeters), and head circumference (centimeters) while on PN therapy. Assessments were made when enteral feeds were changed from breast milk or gestational age–appropriate standard cow's-milk protein–based infant formula to a PH or CAA formula. The institutional review board at the University of Chicago approved the study.

Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at the University of Chicago. REDCap is a secure, web-based application designed to support data capture for research studies, providing an intuitive interface for validated data entry, audit trails for tracking data manipulation and export procedures, automated export procedures or seamless data downloads to common statistical packages, and procedures for importing data from external sources (9). For statistical analysis, the data were presented as actual growth measurements, z scores, medians, ranges, and percentages. Infants who received multiple courses of PN were further analyzed to determine those diagnosed with CMPI based on the requirement for change of feeds to PH or CAA formula. The duration of hospitalization and clinical characteristics of infants diagnosed with CMPI and those without CMPI were compared using Wilcoxon rank-sum tests for continuous variables and χ2 tests for categorical variables. A P < 0.05 was considered significant.


Data were analyzed from 348 infants who were identified as having received PN support therapy during the study period. The median gestational age for this cohort was 32 weeks with a median birth weight of 1618 g (range 425–5510 g). Median start date of PN therapy was on day 1 of life with a median initial course lasting 1.3 weeks (range 0.1–16 weeks). Further clinical characteristics of this cohort are described in Table 1.

Clinical characteristics of study population receiving PN therapy (N = 348)

Fifty-one of 348 infants (15%) received >1 course of PN (42 had 2 courses and 9 had ≥3 courses). Figure 1 displays the method of study subject identification. Nineteen of the 348 (5%) infants on PN received a PH- or CAA-based formula at some point during their hospitalization. The infants who required multiple courses of PN (N = 51) were more likely to have required a PH or CAA formula compared with those who had received only a single course of PN: 27% versus 2%, P < 0.001. Table 2 displays the clinical characteristics of infants who received multiple courses of PN. The clinical characteristics of these 51 infants did not significantly differ based on the likelihood for diagnosis of CMPI; however, those subsequently diagnosed with CMPI were discharged later (19.4 vs 8.3 weeks, P = 0.012), thus associating CMPI with prolonged hospital stay regardless of birth weight.

Flowchart of study population receiving PN therapy (N = 348 patients). CAA = crystalline amino acid formula; NEC = necrotizing enterocolitis; PH = protein hydrolyzed formula; PN = parenteral nutrition; SIP = spontaneous intestinal perforation.
Characteristics of patients requiring multiple courses of PN (N = 51)

Among the group requiring multiple courses of PN therapy and diagnosed with CMPI (N = 14), 9 (64%) infants were diagnosed with NEC at a median age of 28 days (range 22–79) that was preceded by a median duration of 22 days (range 19–55) of enteral feeding with an intact protein. The clinical courses of 14 infants who required multiple courses of PN and eventually diagnosed with CMPI are displayed in Table 3. Most of the infants received intact protein formula alone or in combination with breast milk as first choice of enteral nutrition. Abdominal distension, increased gastric residuals, and emesis were the most common symptoms documented, whereas only 5 patients developed hematochezia.

Characteristics of 14 patients with multiple courses of PN and diagnosed with CMPI

Patients 1 and 5 initially presented with a spontaneous intestinal perforation before the initiation of any enteral nutrition. There were only 2 patients (3 and 14) who received strictly breast milk feeds, although maternal dietary history was unknown. These 2 patients received breast milk for >3 weeks before the onset of a NEC-like episode. Patient 3 was able to transition from CAA to PH formula at the time of discharge without recurrence of symptoms. Patient 12 was started on a soy-based formula for suspicion of galactosemia, which was eventually ruled out but was later transitioned to CAA because of ongoing feeding intolerance.

There was no association between duration of enteral feeds and time to onset of NEC with regard to birth weight. On resumption of enteral feeding with standard infant formula, there was recurrence of feeding intolerance symptoms in all infants. In addition, 4 infants had recurrence of NEC-like illness at a mean age of 66 days that persisted until change of enteral feed to a PH or CAA formula. All of the 14 patients were discharged on a PH or CAA formula with resolution of feeding intolerance symptoms.


This report provides the nutrition support perspective of preterm infants who ultimately were diagnosed with CMPI. They represented 5% of all infants who received PN therapy and the clinical presentations ranged from repeated feeding intolerance symptoms to “late-onset or recurring NEC-like illness” that required multiple courses of PN therapy. We observed a positive response following change of feeds to a PH or CAA formula. These findings demonstrate that CMPI can present in preterm infants and that when correctly diagnosed it is an easily treatable condition. It was also apparent that the diagnosis of CMPI and implementation of appropriate dietary therapy were often delayed in favor of “late onset” or “recurrence” of NEC illness.

CMPI in term infants classically manifests as colic, reflux, diarrhea, hematochezia, or generalized feeding intolerance following a period of antigenic exposure and sensitization that may range from 10 days to 10 months (10). Reports in preterm infants within 24 hours of initial exposure have been observed, suggesting the possibility of intrauterine sensitization in some patients (11). Several investigators have described cases of CMPI in preterm infants while receiving intact milk protein that resolved with its elimination and recurred with its reintroduction. Coviello et al (12), Hirose et al (13), and de Peyer and Walker-Smith (14) have all separately presented cases of hematochezia and enterocolitis, in preterm and newborn infants, that was attributed to CMPI with the resolution of symptoms, without recurrence, after cow's-milk protein was eliminated from their diet. CMPI is a clinical diagnosis that can be confirmed by the resolution of symptoms after the elimination of dairy protein (15).

NEC is a gastrointestinal emergency, therefore appropriately the first concern when preterm or newborn infants present with hematochezia or feeding intolerance. CMPI and NEC are difficult to differentiate because currently no helpful serum or laboratory markers exist to help distinguish one disease from the other. In a retrospective study of 275 neonates presenting with hematochezia, ∼20% also had peripheral eosinophilia; however, the presence of eosinophilia did not predict a more benign course of NEC to suggest that primary diagnosis might have been CMPI (16). Tissue eosinophilia is a more specific indicator of allergic colitis than peripheral eosinophilia; however, invasive testing is not always justified during an acute presentation of hematochezia (13). Similarly, specific immunoglobulin E (IgE) and skin-prick tests are not useful for the diagnosis of CMPI, especially in preterm infants.

The presenting symptoms of CMPI range from feeding intolerance, hematochezia, and pneumatosis intestinalis to acute abdomen and septic shock (17). Feeding intolerance in preterm infants and critically sick newborns presents as prolonged requirement for PN therapy. Improved tolerance of enteral feeds accompanied by more rapid weaning off PN therapy has been observed in infants with short-bowel syndrome who were fed human milk or had enteral feeds changed to PH or CAA formula (18–20). In our study, 8 of the 14 infants who required multiple courses of PN because of persistent feeding intolerance symptoms had been fed a standard Cow's-milk protein–based preterm formula before the initial diagnosis of NEC. Also, 2 infants being fed breast milk developed symptoms diagnosed as NEC, and 2 infants subsequently diagnosed with CMPI presented with spontaneous intestinal perforation before initiation of feeds. After the treatment of NEC and the resumption of standard cow's-milk protein–based preemie infant formula, the infants continued to manifest feeding intolerance symptoms and dependence on PN that resolved only after change of feeds to a PH or CAA formula. This clinical course suggested that events related to severe gastrointestinal injury (intestinal perforation) and the onset of NEC or NEC-like illness may be involved in the pathogenesis of CMPI in preterm infants.

Our findings have limitations that should be considered during the interpretation of this article. The study was retrospective, observational, and diagnosis of CMPI based on resolution of clinical symptoms following empiric change of feeds to a PH or CAA formula. Despite these limitations, the findings of CMPI developing after NEC have validity. Increased β-lactoglobulin, casein-specific interferon-γ and interleukin-4 (IL-4) proinflammatory effector cytokine responses have been detected in infants during the recovery phase from NEC, thus suggesting that NEC was a sensitizing event (21). Further assessment of the effector and regulatory cytokine profile following full recovery from NEC and tolerance of enteral feeds have shown a dramatic decline in β-lactoglobulin, casein-specific interferon-γ, and IL-4–secreting cells accompanied by increased transforming growth factor-β1 (TGF-β1) cells. These findings suggest that full recovery from NEC and tolerance of enteral feeds is accompanied by a switch from largely proinflammatory β-lactoglobulin- and casein-specific cytokines to a profile of predominantly TGF-β regulatory cytokines. Thus, failure of this switch may underlie the onset and persistence of CMPI after NEC and/or severe gastrointestinal injury (21). Furthermore, disruption of the commensal flora by antibiotic therapy initiated during the immediate perinatal and postnatal period may also play a role in inducing sensitization to dietary antigens in preterm and newborn infants. Commensal flora is essential for inducing mucosal IgA and T-regulatory responses that are important for preventing intestinal inflammation (22). Stefka et al (23) have shown that neonatal mice pretreated with antibiotics to disrupt their commensal flora before weaning, then later fed peanut antigens, had significantly reduced ileal bacterial load and diversity, and also developed a hyperreactive IgE response to peanuts compared with nontreated controls, thus linking antibiotic therapy and suppression of the intestinal microbiome to increased risk for food allergen sensitization. Increased serum concentrations of IgE is more likely to be found in newborn infants with a strong maternal history of atopic disease (24). The ability to synthesize IgE is present as early as 11 to 21 weeks of gestation (25). Therefore, preterm infants have the potential to mount an antigen-induced immune response. We speculate that the persistent post-NEC feeding intolerance may represent failure to switch from the proinflammatory effector cytokine state to a TGF-β predominant regulatory phase that is important for the induction of tolerance to dietary antigen. Therefore, NEC may be a sensitizing event for CMPI and that some forms of NEC, especially the recurring or late-onset variant, may in fact be the manifestation of CMPI. Feeding intolerance in preterm infants regardless of etiology presents as prolonged dependence on PN. Therefore, we propose that in situations when human milk is not available, clinicians should exercise a low threshold for changing feeds either to a PH or CAA formula in preterm or sick newborn infants presenting with symptoms of persistent feeding intolerance after NEC, intestinal perforation, recurring NEC and late-onset NEC.

Going forward, there is a need for prospectively designed studies that systematically differentiate NEC from CMPI-related NEC-like illness. There is also a need for standardized ways to diagnose CMPI in preterm infants. Peripheral eosinophilia and allergy testing have been proposed as potential markers for allergic enterocolitis (16). Furthermore, among infants who receive breast milk feeds, there needs to be documentation about maternal dairy intake, other food restrictions during lactation, and family history of atopic diseases. Prospective studies based on assessments of the microbiome and measurements of regulatory- and antigen-specific inflammatory cytokines may help to distinguish or clarify the associations between “NEC,” “NEC-like illness,” and CMPI in preterm infants.

CMPI and NEC appear to have very similar clinical manifestations in preterm and sick newborn infants, and both may lead to the implementation of PN therapy. There is evidence that profound mucosal injury may predispose to sensitization to dietary antigens and result in a syndrome of post-NEC CMPI. Similarly, recurring NEC-like illness may be a clinical manifestation of misdiagnosed CMPI or post-NEC sensitization and onset of CMPI. Therefore, from a nutrition support perspective, we propose that a clinical course characterized by prolonged or multiple courses of PN because of persistent feeding intolerance following NEC, or recurrence of NEC-like illness in preterm infants, should lead to the consideration of CMPI and appropriate management action to change feeds from an intact protein to a PH or CAA formula. Prompt recognition and implementation of dietary change may reduce the duration of hospital stay and morbidity and mortality in this vulnerable population.


1. Pediatric Nutrition Handbook. 6th ed. Elk Grove Village, IL: American Academy of Pediatrics; 2009:519–40.
2. Ludman S, Shah N, Fox AT. Managing cows’-milk allergy in children. BMJ 2013; 347:f5424.
3. Kawai M, Kubota A, Ida S, et al Cow's milk allergy presenting Hirschsprung's disease-mimicking symptoms. Pediatr Surg Int 2005; 21:850–852.
4. Fanaroff AA, Stoll BJ, Wright LL, et al Trends in neonatal morbidity and mortality for very low birthweight infants. Am J Obstetr Gynecol 2007; 196: 147.e1–8.
5. Neu J, Walker WA. Necrotizing enterocolitis. N Engl J Med 2011; 364:255–264.
6. Obladen M. Necrotizing enterocolitis—150 years of fruitless search for the cause. Neonatology 2009; 96:203–210.
7. Fell JM. Neonatal inflammatory intestinal diseases: necrotising enterocolitis and allergic colitis. Early Hum Dev 2005; 81:117–122.
8. Kastenberg ZJ, Sylvester KG. The surgical management of necrotizing enterocolitis. Clin Perinatol 2013; 40:135–148.
9. Harris PA, Taylor R, Thielke R, et al Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Informat 2009; 42:377–381.
10. Martorell A, Plaza AM, Bone J, et al Cow's-milk protein allergy. A multi-centre study: clinical and epidemiological aspects. Allergol Immunopathol (Madr) 2006; 34:46–53.
11. Faber MR, Rieu P, Semmekrot BA, et al Allergic colitis presenting within the first hours of premature life. Acta Paediatr 2005; 94:1514–1515.
12. Coviello C, Rodriquez DC, Cecchi S, et al Different clinical manifestation of cow's-milk allergy in two preterm twins newborns. J Matern Fetal Neonatal Med 2012; 25 (suppl 1):132–133.
13. Hirose R, Yamada T, Hayashida Y. Massive bloody stools in two neonates caused by cow's-milk allergy. Pediatr Surg Int 2006; 22:935–938.
14. de Peyer E, Walker-Smith J. Cow's-milk intolerance presenting as necrotizing enterocolitis. Helv Paediatr Acta 1978; 32:509–515.
15. Dupont C, Soulaines P, Lapillonne A, et al Atopy patch test for early diagnosis of cow's-milk allergy in preterm infants. J Pediatr Gastroenterol Nutr 2010; 50:463–464.
16. Christensen RD, Lambert DK, Gordon PV, et al Neonates presenting with bloody stools and eosinophilia can progress to two different types of necrotizing enterocolitis. J Perinatol 2012; 32:874–879.
17. Berman L, Moss RL. Necrotizing enterocolitis: an update. Semin Fetal Neonatal Med 2011; 16:145–150.
18. Andorsky DJ, Lund DP, Lillehei CW, et al Nutritional and other postoperative management of neonates with short bowel syndrome correlates with clinical outcomes. J Pediatr 2001; 139:27–33.
19. Bines J, Francis D, Hill D. Reducing parenteral requirement in children with short bowel syndrome: impact of an amino acid-based complete infant formula. J Pediatr Gastroenterol Nutr 1998; 26:123–128.
20. Diamanti A, Fiocchi AG, Capriati T, et al Cow's-milk allergy and neonatal short bowel syndrome: comorbidity or true association? Eur J Clin Nutr 2015; 69:102–106.
21. Abdelhamid AE, Chuang SL, Hayes P, et al Evolution of in vitro cow's-milk protein-specific inflammatory and regulatory cytokine responses in preterm infants with necrotising enterocolitis. J Pediatr Gastroenterol Nutr 2013; 56:5–11.
22. Maynard CL, Elson CO, Hatton RD, et al Reciprocal interactions of the intestinal microbiota and immune system. Nature 2012; 489:231–241.
23. Stefka AT, Feehley T, Tripathi P, et al Commensal bacteria protect against food allergen sensitization. Proc Natl Acad Sci USA 2014; 111:13145–13150.
24. Kuiper S, Muris JW, Dompeling E, et al Association between first-degree familial predisposition of asthma and atopy (total IgE) in newborns. Clin Exp Allergy 2006; 36:594–601.
25. Miller DL, Hiravonen T, Gitlin D. Synthesis of IgE by the human conceptus. J Allergy Clin Immunol 1973; 52:182–188.

cow's-milk protein intolerance; milk allergy; necrotizing enterocolitis; prematurity; preterm infants

© 2016 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology,