Digital Image Analysis of Heterogeneous Tuberculosis Pulmonary Pathology in Non-Clinical Animal Models using Deep Convolutional Neural Networks

Efforts to develop effective and safe drugs for treatment of tuberculosis require preclinical evaluation in animal models. Alongside efficacy testing of novel therapies, effects on pulmonary pathology and disease progression are monitored by using histopathology images from these infected animals. To compare the severity of disease across treatment cohorts, pathologists have historically assigned a semi-quantitative histopathology score that may be subjective in terms of their training, experience, and personal bias. Manual histopathology therefore has limitations regarding reproducibility between studies and pathologists, potentially masking successful treatments. This report describes a pathologist-assistive software tool that reduces these user limitations, while providing a rapid, quantitative scoring system for digital histopathology image analysis. The software, called 'Lesion Image Recognition and Analysis' (LIRA), employs convolutional neural networks to classify seven different pathology features, including three different lesion types from pulmonary tissues of the C3HeB/FeJ tuberculosis mouse model. LIRA was developed to improve the efficiency of histopathology analysis for mouse tuberculosis infection models, this approach has also broader applications to other disease models and tissues. The full source code and documentation is available from https://Github.com/TB-imaging/LIRA.

Cow milk feeding in infancy: further observations on blood loss from the gastrointestinal tract

Because feeding of cow milk causes normal infants to lose increased amounts of occult blood from the gastrointestinal tract, we conducted a prospective trial to measure intestinal blood loss quantitatively and to monitor iron nutritional status. Fifty-two infants entered the trial at 168 days of age and were assigned at random to receive either cow milk or a milk-based formula. Initially, 31 infants had been breast-fed and 21 had been fed formulas. With the feeding of cow milk, the proportion of guaiac-positive stools increased from 3.0% at baseline to 30.3% during the first 28 days of the trial (p less than 0.01), whereas the proportion of positive stools remained low (5.0%) with the feeding of formula. The proportion of guaiac-positive stools among cow milk-fed infants declined later, but for the entire trial it remained significantly (p less than 0.01) elevated. Stool hemoglobin concentration increased markedly with the introduction of cow milk, rising from a mean (+/- SD) of 622 +/- 527 micrograms/gm dry stool at baseline to 3598 +/- 10,479 micrograms/gm dry stool during the first 28 days of ingestion of cow milk. Among infants fed formula, stool hemoglobin did not increase and was significantly (p less than 0.01) less than in the cow milk group. Among infants fed cow milk, the increase in hemoglobin concentration tended to be greater for those who had initially been fed human milk than for those who had initially been fed formulas. Iron nutritional status was not significantly different between the two feeding groups. However, one infant became iron deficient after 4 weeks of ingesting cow milk. We conclude that cow milk feeding leads to increased intestinal tract blood loss in a large proportion of normal infants and that the amount of iron lost is nutritionally important.

Effect of low zinc intake on absorption and excretion of zinc by infants studied with 70Zn as extrinsic tag

The effect of low dietary intake of zinc was studied in six normal infants with the use of 70Zn as an extrinsic tag. Of the two study formulas, one provided a zinc intake similar to that of customary infant formulas ("high" intake), whereas the other provided a "low" zinc intake. Two zinc absorption studies were performed with each formula (sequence: high-low-low-high). Extrinsically labeled formula was fed for 24 h and excreta were collected for 72 h. Zinc isotope ratios were determined by inductively coupled plasma mass spectrometry (ICP/MS). When zinc intake was high, net zinc absorption was 9.1 +/- 8.7% (mean +/- SD) of intake and net zinc retention was 74 +/- 91 micrograms/(kg.d). True zinc (70Zn) absorption was 16.8 +/- 5.8% of intake and fecal excretion of endogenous zinc was 78 +/- 56 micrograms/(kg.d). When zinc intake was low, net absorption of zinc increased significantly (P less than 0.001) to 26.0 +/- 13.0% of intake, but net retention was not significantly different at 42 +/- 33 micrograms/(kg.d). True absorption of zinc also increased significantly (P less than 0.001) to 41.1 +/- 7.8% of intake, whereas fecal endogenous zinc decreased (P less than 0.05) to 34 +/- 16 micrograms/(kg.d) during low zinc intake. Thus, infants maintain zinc balance in the face of low zinc intake through increased efficiency of absorption and decreased excretion of endogenous zinc.

Intrinsic and extrinsic stable isotopic zinc absorption by infants from formulas

The kinetics of fecal elimination and the extent of fractional absorption of intrinsic and extrinsic zinc from infant formula were studied in 11 human infants by using a direct, simultaneous dual-label method based on fecal monitoring of unabsorbed stable isotope labels. The relative positions of the two stable isotopic tags (67Zn, 70Zn) as extrinsic and intrinsic labels were reversed in two separate determinations of fractional absorption. Administration of doubly labeled formula for 24 h or less permitted elimination of unabsorbed labels before the end of a 72-h metabolic interval. The extent of enrichment and the kinetics of fecal elimination of unabsorbed extrinsic stable isotopic zinc tag were observed to agree closely with the corresponding values for the intrinsic tag. Fractional absorption of intrinsic and extrinsic tags was highly correlated and did not differ significantly. The ratio of fractional absorption of extrinsic tag to that of intrinsic tag was 1.05 +/- 0.19 (mean +/- SD). The results constitute strong support for validation of the use of enriched stable isotopes at low levels (less than or equal to 10%) of extrinsic addition to milk-based diets for studies of zinc absorption by humans.

Iron absorption from infant foods

To determine the bioavailability of iron from iron-fortified infant foods, we have determined erythrocyte incorporation of the stable isotope, 58Fe, after feeding the following foods extrinsically labeled with 58Fe: 1) rice cereal with apples and bananas ("cereal-fruit product"), 2) Mead Johnson Enriched Baby Food (MJEBF), a vitamin, mineral, and protein-enriched rice cereal, 3) vegetables and beef ("vegetable-beef product"), 4) grape-juice, and 5) MJEBF. Foods 1-4 were fortified with ferrous sulfate, and food 5 was fortified with ferrous fumarate. Blood was obtained at ages 140, 168, and 196 d of age, and the test meal was fed under standardized conditions at 154 d of age. Erythrocyte incorporation of the 58Fe label was determined from the increase in the mass isotope ratio, 58Fe/57Fe, from the baseline value (at 140 d of age) to the follow-up values. The mass isotope ratio was determined by inductively coupled mass spectrometry. Geometric mean total iron incorporation into erythrocytes from the test meal of MJEBF fortified with ferrous sulfate (food 2) was 0.05 mg, and from the vegetable-beef product test meal (food 3) was 0.08 mg. The low value for MJEBF is presumably explained by the low level of iron fortification. The low value for the vegetable-beef product may reflect the presence of inhibitors of iron absorption. Geometric mean erythrocyte incorporations of iron from the test meals with foods 1, 4 and 5 were 0.15, 0.14, and 0.18 mg, respectively. These erythrocyte incorporation values are 20 to 26% of the estimated 0.7 mg requirement for absorbed iron, and therefore seem nutritionally important.

Erythrocyte incorporation of ingested 58-iron by infants

The least abundant stable isotope of iron, 58Fe (natural abundance 0.322 weight %), was administered orally to infants to explore the feasibility of using a stable rather than a radioisotope in studies of iron absorption. The dose of 58Fe was given between feedings at age 126 days. The mass isotope ratio, 58Fe/57Fe, was determined in blood by inductively coupled plasma mass spectroscopy and at ages 140, 168, and 196 days. The percentage of the 58Fe dose entering the circulation (3.2 to 16.0%) was inversely correlated with serum ferritin concentration (r = -0.867, p less than 0.01). For individual infants the SD of the percentage of administered dose of iron appearing in the circulation ranged from 0.22 to 1.28. We conclude that the method is likely to be suitable for within-subject comparisons of iron availability from foods. Because of the large between-subject variation, we are pessimistic for this age group about the usefulness of study designs based on group comparisons.

Lack of adverse reactions to iron-fortified formula

Some physicians are reluctant to recommend feeding of iron-fortified formulas to infants because of a fear of adverse reactions. In crossover studies, parents' records were compared with regard to their infant's behavior (fussiness, cramps, regurgitation, flatus, colic) and stool characteristics during periods when iron-fortified formulas were fed and periods when non-iron-fortified formulas were fed. No statistically significant feeding-related difference was noted except for stool color.

Bioavailability of dietary urea nitrogen in the infant

Because the human body has no enzymes capable of hydrolyzing urea, nitrogen from this source becomes bioavailable only by release of ammonia from urea by bacterial hydrolysis in the intestines, with subsequent absorption and utilization of ammonia. To explore extent to which urea ingested in milk becomes bioavailable, we fed di-15N-urea (both nitrogen atoms in the form of the stable isotope 15N) and determined urinary excretion of di-15N-urea (excreted without having become bioavailable) and mono-15N-urea (urea containing only one atom of 15N and therefore reflecting excretion of absorbed ammonia). The largest percentage of the ingested di-15N-urea was excreted promptly in the urine still in the form of di-15N-urea. We conclude that most of the urea ingested by a normal infant is not bioavailable.

Effect of iron fortification of infant formula on trace mineral absorption

This study was designed to examine whether iron fortification of infant formulas has an effect on utilization of other nutrients, particularly the trace elements zinc and copper. Metabolic balance studies were performed with seven normal infants who were between 43 and 420 days of age. Two formulas of nearly identical composition except for iron concentration (10.2 and 2.5 mg/L) were fed. Each infant had four balance studies performed, two while being fed formula 10.2 and two while being fed formula 2.5, in an alternating sequence. No effect of formula iron concentration was evident on absorption and/or retention of nitrogen, fat, calcium, and magnesium. Although absorption of phosphorus was significantly (p less than 0.05) less with formula 10.2 than formula 2.5, the difference was trivial. No effect on absorption of zinc was seen. However, absorption of copper was only 13.4% (SD 13.0) of intake when formula 10.2 was fed, compared with 27.5% (SD 15.3) of intake when formula 2.5 was fed. The difference was statistically significant (p less than 0.01). We conclude that iron in amounts present in iron-fortified formulas has a measurable effect on copper utilization. Because the magnitude of the effect is relatively small, we doubt that the finding is clinically relevant.

Requirement for sulfur-containing amino acids in infancy

A series of studies designed to define the requirement of normal infants for sulfur-containing amino acids (methionine, cystine) was conducted with formulas providing 3.0, 2.8, 2.6, 2.2 or 1.8 g of isolated soy protein per 100 kcal. The formulas were fed with or without a methionine supplement. Adequacy of the diet was determined by measurement of growth, serum chemical indices and nitrogen balance. Nitrogen balance demonstrated a beneficial effect of methionine supplementation only at the lowest protein concentration (1.8 g/100 kcal). However, measurement of weight gain and/or serum concentrations of urea nitrogen and albumin clearly showed a beneficial effect of methionine supplementation at protein concentrations of 2.2 and 2.6 g/100 kcal. Intakes of sulfur-containing amino acids of 435 and 495 mumol/100 kcal therefore appear inadequate. At higher intakes of protein (2.8 and 3.0 g/100 kcal) there was no beneficial effect of methionine supplementation. Possible exceptions were male infants provided with 3.0 g protein per 100 kcal, in whom weight gain between 8 and 56 d of age was significantly (P less than 0.05) greater with than without a methionine supplement. Based on intakes of sulfur-containing amino acids from the formula providing 2.8 g of isolated soy protein per 100 kcal without methionine supplementation, we conclude that for male infants older than 56 d the requirement for sulfur-containing amino acids is no more than 588 mumol/100 kcal when intake of methionine is 264 mumol/100 kcal. However, it seems possible that such intake fails to meet the requirement in male infants less than 56 d of age. For female infants, regardless of age, 533 mumol/100 of sulfur-containing amino acids per 100 kcal meet the requirement when intake of methionine is 239 mumol/100 kcal.

Sweetness of diet and food consumption by infants

Normal female infants were studied from 8 to 112 days of age to determine the effect on food consumption of feeding formulas that differed widely in sweetness. One formula contained sucrose and the other a bland-tasting cornstarch hydrolysate. In a balanced, crossover design, it was found that food consumption was significantly greater during intervals in which the sweeter formula was fed.

Cow milk feeding in infancy: gastrointestinal blood loss and iron nutritional status

Eighty-one normal infants were studied between 112 and 196 days of age. Thirty-nine infants were fed pasteurized cow milk and the remainder were fed either Enfamil or heat-treated cow milk. During the age interval of 112 to 140 days, the proportion of infants with guaiac-positive stools was significantly (P less than 0.01) greater among infants fed pasteurized cow milk than among those fed Enfamil or heat-treated cow milk. Similarly, infants fed cow milk had a significantly (P less than 0.001) greater number of guaiac-positive stools than did the other infants. After 140 days of age, there was no difference between feeding groups in the number of guaiac-positive stools. No significant differences were observed in mean hemoglobin, hematocrit, serum iron, total iron-binding capacity, or transferrin saturation between feeding groups nor between infants with and those without guaiac-positive stools, It is concluded that pasteurized cow milk should not be fed before 140 days of age.

Methionine fortification of a soy protein formula fed to infants

Data from study of nine normal full-term infants fed a soy isolate-based formula unsupplemented with methionine were compared with similar data from study of 10 similar infants fed the same formula supplemented with L-methionine and with data from previous studies of larger groups of infants receiving various other feedings. Food intake, growth, and serum chemical values were studied from 8 through 111 days of age. In addition, nitrogen balance studies were carried out. Statistically significant differences were as follows: lesser weight gain per 100 kcal by infants fed the unsupplemented soy isolate-based formula than by infants fed milk-based or other soy isolate-based formulas; lesser serum concentrations of albumin at age 28 days by infants fed the unsupplemented soy isolate-based formula than by breast-fed infants; greater serum concentrations of urea nitrogen by infants receiving the unsupplemented soy isolate-based formula than by those receiving the same formula supplemented with L-methionine. A number of other differences was noted but were not statistically significant. The results suggest that normal infants fed a formula providing 2.25 /100 kcal of a soy protein isolate not fortified with methionine performed less well during the first 6 weeks of life than did breast-fed infants and infants fed milk-based formulas or other soy isolate-based formulas fortified with methionine. The limiting nutrient appears to have been methionine.