Growth of alveoli during postnatal development in humans based on stereological estimation

Abstract

Alveolarization in humans and nonhuman primates begins during prenatal development. Advances in stereological counting techniques allow accurate assessment of alveolar number; however, these techniques have not been applied to the developing human lung. Based on the recent American Thoracic Society guidelines for stereology, lungs from human autopsies, ages 2 mo to 15 yr, were fractionated and isometric uniform randomly sampled to count the number of alveoli. The number of alveoli was compared with age, weight, and height as well as growth between right and left lungs. The number of alveoli in the human lung increased exponentially during the first 2 yr of life but continued to increase albeit at a reduced rate through adolescence. Alveolar numbers also correlated with the indirect radial alveolar count technique. Growth curves for human alveolarization were compared using historical data of nonhuman primates and rats. The alveolar growth rate in nonhuman primates was nearly identical to the human growth curve. Rats were significantly different, showing a more pronounced exponential growth during the first 20 days of life. This evidence indicates that the human lung may be more plastic than originally thought, with alveolarization occurring well into adolescence. The first 20 days of life in rats implies a growth curve that may relate more to prenatal growth in humans. The data suggest that nonhuman primates are a better laboratory model for studies of human postnatal lung growth than rats.

alveoli permit gas exchange by providing structure and thin air-surface barriers for the capillary network that result in oxygen diffusion into the blood (4). Despite decades of study, aspects of alveolarization in the human lung are still not fully understood (1, 4, 28–30, 33). They begin forming before birth between weeks 32 and 36 of gestation, and it has been hypothesized that alveolar development continues to roughly 2 yr of age in humans. However, by adulthood, there is a 20-fold increase in gas exchange surface without a concomitant increase in chest size (16). A number of investigators have estimated the number of alveoli vs. age (1, 9, 10, 15, 33). In 1962, Dunnill found that alveoli increase logarithmically from 24 million at birth to 280 million by age 8 yr (10). Thurlbeck indicated that the human lung reaches around 300 million alveoli in adults (33). Ochs et al. reported an estimate of 480 million and that mean alveolar size remains relatively constant, with an age range of 18-41 years of the human subjects (30).

Historically, the human lung has proven difficult to measure morphometrically during development due to obtaining healthy specimens and ethical constraints in acquiring enough samples. Very few reports and a scarcity of samples in each report have given a limited view on the growth of the human lung. New methodology for morphological estimates of the number of alveoli since the previously mentioned reports has been tested and proven accurate and unbiased (20). Whereas many measurements of lung structure can be biased based on tissue fixation, sectioning, sample size, and model assumptions, counting the number of alveoli as outlined in the recently published American Thoracic Society (ATC) standards for quantitative assessment of lung structure may be the best way of assessing growth or comparisons between lungs (18, 37).

Recently, reports have shown that alveolar growth continues from childhood into adolescence in humans and other mammals (17, 19, 23, 29). Our understanding of lung plasticity in the developing human child remains limited based on recent evidence. Alveoli may have the capacity to “catch up” in its growth after insults more easily if the lung does continue to grow throughout adolescence. A report on ozone exposure in postnatal development of rhesus macaques indicates that, after insult, alveolar growth is stunted, but given a time to recuperate, alveolar growth returns to normal (2). The clinical implications that the lung may be more plastic than original thought allows for hope that the lung can recover from damage inflicted early in life or even adulthood.

In this paper, we present stereological data pertaining to the alveolar growth of the human lung from 1 mo to 15 yr of age. The stereological techniques used are design-based and adhere closely to the guidelines published by the ATC (18). This is the first time these stereological techniques have been used to estimate human alveolar numbers based on age and may represent the most accurate estimate of alveolar growth in humans to date.

MATERIALS AND METHODS

Lung tissue was obtained from autopsies performed at the Sacramento County Coroner's Office. All specimens were coded, and identifiers were not available to the research personnel. Pertinent clinical data, final autopsy findings, and cause of death were available for all cases. The study was approved by the Research Review Committee of the Sacramento County Coroner's Office and determined to be exempt by the University of California, Davis, Committee on Human Research.

Lungs removed at the time of postmortem examination were weighed, and a cannula was inserted in each mainstem bronchus. The cannulas were secured, and the lungs were inflated with 10% neutral buffered formalin (25–30 cm fluid pressure; 15–30 min). As lungs inflated to full size, they were suspended in a container containing identical fixative. Next, the lungs were removed from fixative, and their displacement volume was determined by measuring each lung's buoyant weight in water (average of 3 measurements). With the use of a specially designed cutting board, each lung was cut into 10-mm-thick sagittal slices. Forensic pathologists examined the slices, and, if no gross lesions were identified, each lung was analyzed.

With a disposable knife blade and aided by laminated graph paper grid, strips (10 mm wide) were cut from the lungs. Each strip was lined up according to tissue surface size, largest to smallest. The total number of strips was counted and recorded. To form a bell-shaped curve with the bars of tissue (smooth fractionator), every other bar was selected and arranged in a horseshoe-shaped ring of bars (19). Working from the smallest tissue at one end of the right to the other end, a first fraction was chosen. The size of this first fraction was about 10 long strips. Sampling interval (rounded down to the whole number) was determined by dividing the total number of strips by the desired number of bars to be sampled. A random start number was chosen between one and that whole number. Starting at that random start, the first bar was selected, and counting every fraction thereof (by the whole number obtained), selected bars are chosen. Next, the actual number of bars obtained from the sampling was counted.

To calculate the second fraction (f2), the sampled bars were cut into smaller lengths to form 1.5-cm tissue bricks. The bricks are sorted in the same manner as the tissue strips above, with the largest cut tissue surface up to the smallest, selecting out every other brick and arranging them in a horseshoe shape to obtain a smooth fractionator. Next, tissue bricks are selected for final embedding (f2) in the same manner as before, counting total number of bricks and desired number of bricks equal to eight, dividing the total number of bricks by eight, and selecting a random start number between one and that whole number; then the final tissue to be sampled is selected. The actual number of bricks sampled was determined, and if not the desired number of eight bricks, then the fraction (f2) was recalculated with the correct number of bricks selected. This method represents the isotropic uniform random sampling method that is not required for alveolar number estimation but is required for estimates of surface and length of other anisotropic features in the lung (e.g., airways and vessels).

Each selected brick was placed with the correct tissue surface to be sectioned facing up in the embedding cassettes. A reference block with perfect dimensions (1.5 × 1.0 × 1.0 mm) was embedded in block 1. The selected tissue samples were embedded in paraffin and sectioned 5 μm serially, two sections per slide, two slides total. Sections were routinely stained with hematoxylin and eosin.

Morphological estimates.

Alveolar estimates were performed as previously described (18–20). Briefly, alveoli were identified by alveolar openings into alveolar ducts. Counting the number of alveolar entrance rings in paired sections by the disector technique allows estimation of the total number of alveoli in the lung. The dissector technique uses two images with a known height. In this study, the height was 10 μm. The disector height was a portion of the fraction that was established by fractionating the lung as previously described. This fractionator principle was used in combination with the Eular characteristic of counting alveolar open rings to determine the total number of alveoli in the lung.

Radial alveolar counts.

Radial alveolar counts were estimated to be used as a comparison with the unbiased alveolar counts established by the ATS guidelines. The radial alveolar counts were obtained using Cooney and Thurlbeck's modifications of the original Emery and Mithal method (8, 12). Briefly, radial alveolar counts are obtained by drawing a perpendicular line from the terminal respiratory bronchiole to the nearest and definitive alveolar septal wall. The number of alveoli transected by this line is counted and averaged.

Statistics.

Statistical analyses were based on linear and nonlinear regression (SPSS, IMB, Armonk, NY). Age, body length, body weight, percent of life lived based on life expectancy, and log lung volume were used as predictors against the measured variables, and a series of models were fitted to each. Three functions were considered for each outcome: linear, two-parameter power, and three-parameter power. The models fitted to each outcome were

$$Linear:\hspace{0.17em}y=\hspace{0.17em}{y}^0+ax$$

$$2-parameter\hspace{0.17em}power:\hspace{0.17em}y=\hspace{0.17em}{ax}^b$$

$$3-parameter\hspace{0.17em}power:\hspace{0.17em}y=\hspace{0.17em}{y}^0+{ax}^b$$

Adjusted (Adj) R² value (coefficient of determination) was used to determine the best model fit. The test for R² is whether it is statistically significant from zero. If the model is linear, it will decrease/increase in a uniform fashion. If it is either quadratic or exponential, it will change at a given moment indicating specific inflection points along the curve. Values for y⁰, a, and b are listed in Table 1 for all measured parameters.

Table 1.

Assessment of best-fit model based on age and lung volume as predictors

Species	Variable	Predictor	Model	y0	a	b	Adj R2
Human	Log Nalv,lung	Age, yr	2-Parameter power		8.2551	0.018	0.7502
	Log Nalv (right lung)	Age, yr	2-Parameter power		8.142	0.0148	0.8096
	Log Nalv (left lung)	Age, yr	2-Parameter power		8.0626	0.0172	0.8617
	RAC	Age, yr	2-Parameter power		7.5555	0.2103	0.9254
	Log Nalv,lung	Length	Linear	7.8931	0.0056		0.7767
	Log Nalv,lung	Weight	2-Parameter power		7.851	0.0273	0.8051
	Log Nalv,lung	% of life lived	2-Parameter power		8.989	0.0154	0.8715
	Log Lv	% of life lived	Linear	2.4688	4.0718		0.777
	Log Nalv,lung	Log Lv	Linear	6.0337	0.8921		0.8627
Rhesus	Log Nalv,lung	% of life lived	3-Parameter power	7.8553	1.6701	0.5157	0.7087
	Log Lv	% of life lived	Linear	1.8382	4.7052		0.8249
	Log Nalv,lung	Log Lv	Linear	6.49	0.8098		0.8118
Rat	Log Nalv,lung	% of life lived	3-Parameter power	6.321	4.7706	0.4077	0.7371
	Log Lv	% of life lived	Linear	−0.1966	20.001		0.9262
	Log Nalv,lung	Log Lv	Linear	6.98	0.9011		0.6985

N_alv,lung, no. of alveoli in the lung; N_alv; no. of alveoli; RAC, radial alveolar count; Lv, lung volume.

RESULTS

Study population.

Table 2 lists the sex, age, growth data, and cause of death of study subjects.

Table 2.

Sex, age, growth data, and cause of death of study subjects

Sex	Age	Length, cm	Length, percentile1	Weight, kg	Weight, percentile1	Gestation, wk	Birth Wt, kg	Birth Wt, percentile1	Cause of Death
M	1 mo 3 days	56	75	4.6	61	39.1	3.6	91	SUID while cosleeping
M	1 mo 25 days	57	35	4.6	13	40	3.1	61	SIDS
M	2 mo 9 days	52	<1	3.9	<1	33.4	U	25	CMV myocarditis
M	2 mo 15 days	54	<1	5.0	7	Unknown	U	N/A	SUID while cosleeping
M	3 mo 6 days	66	98	7.3	85	41.6	4.6	98	SUID while cosleeping
M	4 mo 5 days	64.5	56	6.6	25	40	3.9	94	SIDS
F	4 mo 16 days	62	30	6.2	29	36	U	N/A	SUID while cosleeping
F	5 mo	68.5	90	7.7	83	40.4	3.3	65	SIDS
M	2 yr 6 mo	96.5	90	14.1	65	Unknown	U	N/A	Drowning
F	15 yr	165	69	65.3	86	NR	NR	N/A	Blunt force injury of head
M	15 yr 11 mo	150	<1	53.5	22	NR	NR	N/A	Multiple blunt force injuries

M, male; F, female; NR, not relevant; N/A, not applicable; SIDS, sudden infant death syndrome; SUID, sudden unexpected infant death; CMV, cytomegalovirus.

¹For subjects less than age 2, growth percentiles are from World Health Organization growth charts.

For subjects age 2 or older, Centers for Disease Control (CDC) growth percentiles are from CDC growth charts.

There were 11 subjects (3 female; 8 male) ranging in age from 1 mo and 3 days to 15 yr and 11 mo. Five individuals were either below 50% for length or weight for their age based on the Center for Disease Control growth charts. At least two of these individuals were born prematurely. A correction for gestational age was not applied to the growth curves. Death during the first 5 mo of life was primarily due to sudden infant death syndrome or sudden unexpected infant death while cosleeping with a parent. No pulmonary lesions or disorders were noted at autopsy.

Number of alveoli.

Raw values are represented in Table 3 for age, displaced lung volume, and number of alveoli in whole human lung. The log of the number of alveoli in the human lung showed a two-parameter power function curve (Adj R² = 0.75) (Fig. 1 and Table 1). This curve represents a fast increase in growth during the first 2 yr of life that begins to taper off in adolescence but still continues to grow. Right and left lungs were counted separately and mimicked similar growth curves in a two-parameter function (Fig. 2 and Table 1). Left lungs were slightly smaller with a minimal decrease in growth (Adj R² = 0.86) compared with the right lung (Adj R² = 0.81). The log of the number of alveoli also showed a strong correlation to length (Adj R² = 0.78) in a linear function and weight (Adj R² = 0.81) in a two-parameter power function (Figs. 3 and 4, respectively, and Table 1).

Table 3.

Age, sex, DLV, and number of alveoli for whole human lung

Age	Sex	DLV	Nalv,lung
1 mo 3 days	M	238	106514687.5
1 mo 25 days	M	233	143240259.7
2 mo 9 days	M	189	136367142.9
2 mo 15 days	M	240	150306428.6
3 mo 6 days	M	448	204779571.4
4 mo 5 days	M	463	273611688.3
4 mo 16 days	F	318	188067551
5 mo	F	254	161093571.4
2 yr 6 mo	M	735	430630633.9
15 yr 11 mo	M	1360	486792250
15 yy	F	2417	583899357.1

DLV, displaced lung volume.

Fig. 1.

The log of the number of alveoli in the lung (N_alv,lung) vs. age (yr) is plotted according to a 2-parameter power function (Table 1).

Fig. 2.

The log of the number of alveoli in the right lung (circles and solid line) and the left lung (triangles and broken line) vs. age (yr) is plotted according to a 2-parameter power function (Table 1).

Fig. 3.

Log N_alv,lung vs. length (cm) is plotted according to a linear function (Table 1).

Fig. 4.

N_alv,lung vs. weight (kg) is plotted according to a 2-parameter power function (Table 1).

Radial alveolar counts.

Results of the radial alveolar counts correlated strongly with chronological age of the subjects (Adj R² = 0.93). The counts also followed a similar pattern of growth as the number of alveoli in a two-paramater power function (Fig. 5).

Fig. 5.

Radial alveolar counts vs. age (yr) are plotted according to a 2-parameter power function (Table 1).

Comparison with laboratory animals.

Previously reported data from several studies were gathered to compare alveolar growth between humans and two common laboratory animals (Macaca mulatta and Rattus norvegicus). Rodent data were limited to the Sprague-Dawley strain. Lung volume and alveoli data are the averages for that time point represented in previous papers (3, 24, 25). Due to the differences in lifespans of all three species, a comparison was done based on the average life expectancy for each species and the amount of time the subjects lived. For rodents, the average lifespan used was 2.5 yr. Rhesus macaques have an average lifespan of 25 yr, and humans have an average lifespan of 78 yr. The lifespans were converted into days, and this number was divided by the age at necropsy to get a percentage of life lived based on life expectancy. The log of the number of alveoli in humans vs. the percentage of life lived follows a two-parameter power function (Adj R² = 0.87). Rhesus and rats follow a three-parameter power curve (Adj R² = 0.70 and 0.74, respectively) (Fig. 6 and Table 1).

Fig. 6.

N_alv,lung vs. percentage of life lived based on life expectancy for the species for human (circles and solid line), rhesus (triangles and dotted line), and rat (stars and dotted-broken line) are plotted according to a 2- or 3-parameter function (Table 1). Rhesus data were based on previous reports (14, 19) and rat alveoli numbers (3, 24, 25).

Lung volume of each species was also plotted against both the percentage of life lived and the number of alveoli (Figs. 7 and 8 and Table 1). In both instances, all three species followed a linear curve. When assessed against percentage of life lived, lung volume of humans had an Adj R² = 0.78, rhesus had an Adj R² = 0.82, and rodents had an Adj R² = 0.93. Lung volume plotted against the log of number of alveoli had Adj R² = 0.86 for humans, Adj R² = 0.81 for rhesus, and Adj R² = 0.70 for rodents.

Fig. 7.

The log of the lung volume (Log Lv) vs. percentage of life lived based on life expectancy for the species for human (circles and solid line), rhesus (triangles and dotted line), and rat (stars and dotted-dashed line) are plotted according to a 2- or 3-parameter function (Table 1). Rhesus data were based on previous reports (14, 19) and rat alveoli numbers (3, 24, 25).

Fig. 8.

Lung log Lv vs. the log of the number of alveoli in the lung (Log N_alv,lung) for human (circles and solid line), rhesus (triangles and broken line), and rat (stars and dotted-dashed line) are plotted according to a linear function (Table 1). Rhesus data were based on previous reports (14, 19) and rat alveoli numbers (3, 24, 25).

DISCUSSION

Understanding lung structure remains vital to understanding the impact on function either from pathological disturbances or environmental insults for comparative physiology (37). Alveolar number estimation done previously on humans used single independent sections for counting and assumption of a specific geometric shape (33, 36). Other approaches included a selector method of estimating the alveolar volume and dividing that into the volume density of alveoli per lung or reconstructing the total number from a small number of alveoli per lung (18, 24, 27). The disector was introduced in 1984 that allowed for number and size to be counted without the need for assumptions about size, shape, or orientation (32). Ochs et al. used this technique along with the methods outlined in the recently published guidelines for morphometric analysis on lungs to study the number of alveoli in humans (18, 31). This method uses either a fractionator or stratified random sampling and lung volume and the Eular characteristic of the net of alveolar openings to estimate the total number of alveoli in the lung. The fractionator method also removes any bias from lung inflation as well as model assumptions (20). Ochs was limited to human adults with a sample size (2 males and 4 females). However, the data did provide reference values for the adult human lung. The estimates of number of alveoli in the adult human lung ranged from 274 to 790 million in this study, which mirrors the upper range of the 15-yr-old male in our study (31).

Alveolar growth in humans begins 32 wk into gestation and until recently was believed to continue to 2 yr of age with a six- to eightfold increase in alveoli from birth (7, 11). At birth, the alveolar estimate is between 20 and 50 million (22). Our study indicates alveolar numbers at 1 mo to be around 100 million (Fig. 1). Recent evidence from helium-3 magnetic resonance has shown that alveolarization continues during adolescence in humans (29). At the end of somatic growth, alveolar numbers were previously reported to range between 300 and 600 million in humans (31, 38). Evidence presented in this study shows that the range can be higher. The two 15-yr-old subjects (one male and one female) had roughly 650–700 million alveoli (Fig. 1). Figure 2 shows that the right and left lung grow in the same pattern albeit with the left lung slightly smaller.

Lung growth is not linear. Thurlbeck and Angus showed that the volume of the lung grows rapidly during the first 2 yr of life and then stabilizes at around 8 yr (34). Cooney and Thurlbeck reported similar results when performing radial alveolar counts (8). Our findings follow a similar pattern as previously reported using the same methods as well as an analogous growth curve to the number of alveoli vs. age. The correlation to age for radial alveolar counts is stronger than the number of alveoli; however, the counts give different types of data. Radial alveolar counts assess the complexity of the acinus or the amount of septal intersections from a straight line drawn from the respiratory bronchiole to the acinus (8). It remains a quick counting method to generate patterns of growth but remains limited based on lung inflation by fixation. The number of alveoli using a smooth fractionator is not dependent on lung inflation by variable fixation pressures and is solely estimated by the count of alveolar openings and the sampling fraction (19, 20).

A recent review noted that the only hope for patients with dysfunctional lungs due to disordered growth or acquired disease is transplantation (13). If the lung has the capacity to grow throughout adolescence, its ability to regenerate after disease and return some functionality after insult may still exist. Rhesus macaques have been show to return to normal lung growth patterns after an ozone and allergen insult (2). Growth retardation due to protein restriction in rats limited lung volume growth but was capable of recovery to a normal relationship to control groups after refeeding (21). Similarly, alveolarization has been shown to take place in an adult human postpneumonectomy via helium-3 magnetic resonance imaging scanning (5).

Previous estimates of the number of alveoli in the lungs of rhesus macaques were done using the same methodology (14, 19, 20). The values are reported in Fig. 6 alongside the human data. Nonhuman primates are routinely used as a model of human lung diseases. At birth, comparative morphology between macaques and humans indicates similar segmental arrangement, structure and branching of airways, arterial structure, and arterial changes after birth (26, 35). It has been proposed that alveolar growth in rhesus mimics that of humans (19). Figure 6 represents the first time that alveolar growth has been compared between rhesus macaques and humans using a design-based stereological approach. Growth in both species follows a similar pattern in the age range represented. Although humans have a more rapid increase in alveolar number in the first 2 mo, it levels off afterward and continues in the same trajectory as the rhesus.

In a newborn rat, the lung is at the saccular stage of lung development and remains absent of any double capillary walled secondary septa that mark alveolar septal growth (4). The growth of alveoli in the first 60 days of a rat is at a much steeper slope than humans or rhesus as seen in Fig. 6. At day 1, the estimate is around only 1,000,000 alveoli and may be limited in that the rat is still in the saccular stage of growth. The number of alveoli in the rat lung may increase 60 times by day 60 from birth (24). A potential comparison is that prenatal human lung has been shown to have a steeper alveolar growth curve during gestation than after birth (15). The newborn rat may mimic prenatal alveolar development more than postnatal development at least during the first 20–40 days in the rat until growth slows to a rate similar to that seen in rhesus and human data. Differences between rodents and humans also exist in the growth of lung volume (Fig. 7). When matched against the number of alveoli, the slope of all three species is rather similar. This may indicate that, regardless of species, alveolar growth remains fixed around the rate of lung volume expansion (Fig. 8). A limitation in the comparison between species is the lack of distinguishable gender differences, primarily because of the lack of sample size in humans. Rhesus macaques and rodents show similar results, with females having smaller and a higher number of alveoli along with increased alveolar surface area per body weight compared with males (19). A more in-depth study on gender differences in rodents and humans regarding lung growth, alveolar number, and body weight would eliminate this limitation.

Appropriate models of human lung development are required to continue advancing our understanding of lung biology and human medicine. This study indicates that human lung alveolar growth does not conclude between 2 and 3 yr of age. The number of alveoli in our oldest subjects at 15 yr of age is greater than the 2-yr-old by 150 million. Alveolar growth does continue during adolescence but at a reduced rate after 2 yr. The impact of continuing alveolar growth in adolescence may allow for increased lung plasticity and recovery from early childhood lung diseases. Previously, the overall structure of the monkey lung has been shown to be more similar to human lungs than other laboratory animals (35). This study indicates that growth of lung alveoli of rhesus macaques is also more comparable to human lungs than rodents.

GRANTS

This work was supported by National Institutes of Health Grants P01-ES-0628 and OD-011107.

DISCLOSURES

No conflicts of interest, financial or otherwise are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: M.J.H., L.F.P., G.W., and W.E.F. performed experiments; M.J.H. and D.M.H. analyzed data; M.J.H. interpreted results of experiments; M.J.H. prepared figures; M.J.H. drafted manuscript; M.J.H., W.E.F., and D.M.H. edited and revised manuscript; W.E.F. and D.M.H. conception and design of research; W.E.F. and D.M.H. approved final version of manuscript.