Shortness of Breath During Pregnancy: Could a Cardiac Factor Be Involved?

Abstract

Background

Shortness of breath (SOB) is common among healthy women with normal pregnancies. However, when no overt cardiac or extra cardiac etiology is found, a subtle cardiac source must be excluded.

Hypothesis

Pregnancy may induce or unmask myocardial dysfunction that may cause SOB.

Methods

Healthy pregnant women with significant SOB were recruited for this study. We performed a comprehensive echocardiographic assessment including tissue Doppler imaging (TDI) and 2‐ dimensional strain imaging (2DS). The echocardiographic data obtained were compared with that of a control group of pregnant women without SOB.

Results

Thirty pregnant women with SOB were enrolled in the study (age, 31.8 ± 4.9 years, and gestation, 38.2 ± 2.8 weeks) for whom no overt etiology for SOB was detected. Patients with SOB compared with controls had thicker hearts (septum: 10.1 ± 1.1 vs 8.9 ± 0.9 mm; P < 0.001; posterior wall: 9.4 ± 1.1 vs 8.9 ± 0.9 mm; P < 0.01), shorter E‐wave deceleration time (158.0 ± 50.1 vs 187.1 ± 37.6 msec; P = 0.01), and higher pulmonary artery pressure (26.8 ± 6.2 vs 19.0 ± 6.5 mm Hg, P < 0.01). Women with SOB tended to have a lower S' velocity TDI (P = 0.05) and a trend toward increased torsion on 2DS (P = 0.09).

Conclusions

Significant SOB during otherwise normal pregnancy is associated with significant echocardiographic findings that may suggest a subtle cardiac involvement. Further investigation is necessary to verify such an association, which may have therapeutic implications for treating SOB of pregnancy.

Introduction

Shortness of breath (SOB) is common during pregnancy, occurring in 60% to 70% of healthy pregnant women,1 and is considered a normal physiologic response to pregnancy,2, 3, 4 although it may occasionally be a sign of underlying heart or lung disorders. SOB of pregnancy may occur early in pregnancy. It often improves or stabilizes as term approaches,4 and it typically does not interfere with daily activities.2, 4 However, many pregnant women are admitted to hospital for severe SOB and exercise intolerance. They undergo multiple tests to exclude significant cardiac or respiratory disorders, which, in most cases, are not found. There is no therapy for SOB if no underlying treatable cause is identified.

The mechanism of SOB of pregnancy is controversial. Various etiologies have been suggested, including an increased mechanical compression of the lungs, displacement of the diaphragm by the gravid uterus,5, 6, 7 a change in perception of normal respiration, hyperventilation in response to a reduced diffusion lung capacity,6, 7 a higher sensitivity of the central chemo reflex response to carbon dioxide (CO₂),8 and an effect of gestational hormones (progesterone and estradiol) on the drive to breath.9 However, all of the above studies focused only on respiratory causes for SOB and did not consider a possible role of subtle cardiac factors.

The physiologic changes in pregnancy include an increase in blood volume, a decrease in peripheral vascular resistance during the first weeks of pregnancy, and an increase in heart rate (HR).10, 11 These result in an increase in systolic function, notably an increase in cardiac output, a rise in preload, a decrease in afterload, and an increase in left ventricular (LV) mass. A slight increase in LV ejection fraction (LVEF) and LV cavity dimensions have been reported.11 Few data are available on cardiac diastolic function during normal pregnancy,12, 13 and it is not considered to have an impact of clinical significance.

Two recent developments in echocardiography enable a more accurate assessment of diastolic function: a relatively preload‐insensitive estimation of left atrial pressure using tissue Doppler imaging (TDI) in conjunction with standard Doppler14 and a technique based on 2‐dimensional (2D) speckle tracking for quantification of myocardial strain, which provides data on longitudinal and circumferential myocardial function and rotation.15 These deformation indexes were reported to be very sensitive to assess early changes in LV function.16, 17 Thus, recently, a small decrease in LV segmental longitudinal systolic strain in late pregnancy has been described.18

We hypothesized that pregnancy may induce or unmask myocardial dysfunction that may cause SOB. Thus, we attempted to verify if subtle cardiac functional changes could be identified that may have a causative role in the development of significant SOB during otherwise normal pregnancies. For that purpose, we compared the LV myocardial systolic and diastolic function of pregnant women with and without significant SOB using contemporary echocardiographic techniques.

Methods

From January to December 2013, we prospectively studied healthy pregnant women from 2 institutions (Hillel Yaffe Medical Center and Kaplan Medical Center) with no confounding diseases who developed significant SOB and exercise intolerance requiring an emergency department (ED) visit, with or without a subsequent hospital admission. Patients willing to be enrolled in the study underwent an extensive workup for SOB, including detailed history and physical examination, oxygen (O₂) saturation, electrocardiography, Holter monitoring, a dyspnea‐limited treadmill stress test using the modified Bruce or Naughton protocols, pulmonary function tests, routine blood tests (such as complete blood count and blood chemistry, thyroid‐stimulating hormone [TSH], D‐dimer), and a comprehensive echocardiographic study. Chest X‐ray and CT were performed when clinically required.

Women were excluded if they had any structural heart disorder, any systemic disorder, preeclampsia or eclampsia, pulmonary disease, any significant arrhythmia, or a clear cause for SOB found during evaluation.

Conventional echocardiography and Doppler analysis were performed in all patients with SOB and in a control group of pregnant women without SOB matched for age, parity, gravidity, and gestational age. Left ventricular dimensions were measured according to the joint recommendations of the American Society of Echocardiography and European Association of Echocardiography.

Doppler Imaging

Pulsed‐wave Doppler recordings of the mitral inflow were acquired from the apical 4‐chamber view with the sample volume placed at the tips of the mitral valve leaflets. The following parameters were measured by pulsed‐wave Doppler: peak velocities of early (E) and late (A) diastolic filling and E‐deceleration time (DT). Pulsed‐wave TDI was performed from the apical 4‐chamber view using spectral pulsed Doppler signal filters. In the apical 4‐chamber view, a 2‐mm pulsed Doppler sample volume was placed at the level of the medial aspect of the mitral annulus and gain and filter settings were adjusted to optimize the image. High temporal resolution (>100 frames/sec) and a sweep speed of 100 mm/sec were used. Early (E′), late (A′), and diastolic and systolic (S′) annular velocities were measured.

Raw data were stored digitally as DICOM cine loops and transferred for offline analysis to a customized dedicated workstation equipped with custom‐built software (EchoPAC PC Dimension, version 5.0.1; GE Vingmed, Horten, Norway).

Speckle‐Tracking Strain Analysis

Measurements of 2D strain (2DS) and strain rate (2DSR) were obtained by offline semiautomated speckle‐tracking analysis of 2D images of the LV myocardium, recorded from the standard 3 apical and short‐axis views at a frame rate >50 Hz. The endocardial border was semiautomatically traced, and a myocardial region of interest was then automatically identified by the software package. Longitudinal 2DS (%) and 2DSR (strain per second) were measured at peak and end systole in each apical view, and global longitudinal peak strain was obtained by averaging peak 2DS from all 3 views by the EchoPAC software. To determine circumferential 2DS and 2DSR, systolic 2DS and 2DSR were measured in the parasternal short‐axis views at the papillary muscle level. Myocardial rotation in the parasternal short‐axis view was measured at the mitral valve, papillary muscle, and apical levels, from which LV torsion (LV rotation normalized for LV diastolic length) was calculated using the EchoPAC's automated function.

Echo measurements from the group with SOB were compared with those of the matched control group without SOB who underwent a similar echocardiographic study for research purposes. The protocol was approved by the human research institutional review board of both institutions. All patients signed an informed consent form prior to performing any of the above studies.

Statistical Analysis

Mean values and SDs are presented and a 2‐way unpaired t test was performed to determine the statistical significance of differences between groups. P < 0.05 was considered significant.

Results

Thirty pregnant women with significant SOB requiring a visit to an ED and 30 paired controls (healthy pregnant women without SOB) who were routinely followed up in the outpatient clinics of Kaplan (n = 39) and Hillel Yaffe (n = 21) medical centers participated in this study. Baseline characteristics of women with SOB: mean age was 31.8 ± 4.9 years, they presented in their 38.2 ± 2.8 week of gestation, 10 (33%) were primiparas, and 4 (13%) had twins. New York Heart Association class was determined in 83% of the women in the SOB group, and all were ≥ class III (32% were in class III–IV or IV). Mean hemoglobin level was 11.4 ± 1.2 g%; TSH, 1.8 ± 0.8 unit/L; and D‐dimer, 1.0 ± 0.4 mcg/mL. The majority of women were admitted to the hospital for evaluation, and 5 patients underwent evaluation in the ED. The mean heart rate (HR) at rest was 88.2 ± 11.6 bpm; systolic blood pressure (BP), 110.0 ± 13.3 mm Hg; diastolic BP, 66.3 ± 9.9 mm Hg; and O₂ saturation, 98% ± 1%. All patients were in sinus rhythm, and electrocardiography did not reveal any significant abnormalities. The routine blood tests, such as complete blood count, blood chemistry, and TSH, were in normal ranges in all patients.

On Holter monitoring, the mean HR was 88 ± 0.7 bpm (range, 71.6 ± 72–119.7 ± 13.9 bpm). Runs of sinus tachycardia >120/min were found in 64% of women, and 2 patients had short runs of non–clinically significant supraventricular tachycardia. On dyspnea‐limited exercise tests, a significant increase in HR from 102.9 ± 11.5 to 128.6 ± 13.9 bpm (P < 0.001) was found and BP increased from a mean of 106/63 to 124/69 mm Hg (P = 0.03). Exercise time was 4.8 ± 2.5 minutes, and only 4.7 ± 2.8 METs were achieved (30% could complete only 1 stage of Bruce). There was a mild decrease in O₂ saturation during exercise, from 98.2% to 97.7% (in 1 patient to 94%). Nineteen patients with SOB underwent pulmonary function testing that did not reveal significant abnormalities (mean forced expiratory volume at 1 sec/forced vital capacity, 78.4% ± 27.9%; forced expiratory flow 25%–75%, 3.4 ± 0.9).

An additional evaluation in cases with severe SOB was done to exclude serious pathology such as pulmonary embolism (Doppler of lower extremities to exclude deep‐vein thrombosis in 10 patients, computed tomography in 6 patients, and chest X‐ray in 16 patients were normal).

The control group included 30 matched healthy pregnant women without dyspnea. Table 1 shows similar clinical characteristics of patients in both groups, except for a higher HR in patients with SOB (88 ± 12 vs 79 ± 13 bpm; P < 0.01).

Table 1.

Comparison of Obstetric Characteristics Between Pregnant Women With SOB and Controls

	SOB Group, n = 30	Control Group, n = 30	P Value
Maternal age, y	31.8 ± 4.9	30.7 ± 4.0	NS
Gestational age, wk	38.2 ± 2.8	38.6 ± 1.7	NS
Gravida (no. of pregnancies)	2.9 ± 2.1	2.6 ± 1.5	NS
Maternal, kg	63.3 ± 17.3	75.4 ± 16.4	NS
Hemoglobin, g%	11.4 ± 1.2	11.7 ± 1.1	NS
Newborn weight, g	3567 ± 630	3534 ± 707	NS
Apgar score, 1 min	9.0 ± 0.2	9.1 ± 0.2	NS
Apgar score, 5 min	10.0 ± 0.0	10.0 ± 0.0	NS

Abbreviations: NS, not significant; SOB, shortness of breath.

Echocardiography

Table 2 presents a comparison of standard 2‐dimensional echocardiography, TDI, and 2DS imaging results between the group of pregnant women with SOB and the control group. No significant differences were found in LV or left atrial size, degree of mitral regurgitation, or in LVEF. However, patients with SOB had thicker hearts compared with the controls (interventricular septum: 10.1 ± 1.1 vs 8.9 ± 0.9 mm, P = 0.0002; posterior wall: 9.4 ± 1.1 vs 8.5 ± 1.2 mm, P = 0.003; relative wall thickness: 0.41 ± 0.07 vs 0.37 ± 0.05 mm, P = 0.03), along with a trend toward a somewhat lower LV outflow tract velocity time interval (17.1 ± 3.9 vs 22.3 ± 5.4 cm; P = 0.07). On Doppler of mitral inflow, a shorter DT was found in patients with SOB (158.0 ± 50.1 vs 187.1 ± 37.6 sec; P = 0.01) and elevated estimated pulmonary artery pressures (26.8 ± 6.2 vs 19.0 ± 6.5 mm Hg; P = 0.0007) compared with controls were noted (Figure 1). No significant differences in TDI velocities were obtained between the groups.

Table 2.

Comparison of Echocardiographic Measures Between Pregnant Women With SOB and Controls

	SOB Group, n = 30	Control Group, n = 30	P Value
Conventional echo
LVEDd, mm	45.2 ± 4.5	45.3 ± 3.9	NS
LVESd, mm	29 ± 4.6	30 ± 4.4	NS
IVS, mm	10.1 ± 1.1	8.9 ± 0.9	0.0002
PW, mm	9.4 ± 1.1	8.5 ± 1.2	0.003
RWT	0.41 ± 0.07	0.37 ± 0.05	0.03
LA, mm	30.7 ± 5.1	32.4 ± 4.6	NS
LA area, mm2	16.7 ± 2.8	18.6 ± 4.6	0.07
LVEF, %	59.2 ± 2.8	59.6 ± 10.6	NS
Pulmonary pressure, mm Hg	26.8 ± 6.2	19.0 ± 6.5	0.0007
LVOT VTI, cm	17.1 ± 3.9	22.3 ± 5.4	0.07
E, cm/sec	85.9 ± 24.8	83.2 ± 17.6	NS
A, cm/sec	60.4 ± 14	58.7 ± 13.2	NS
E/A	1.4 ± 0.3	1.5 ± 0.5	NS
DT, msec	158 ± 50.1	187.1 ± 37.6	0.01
Tissue Doppler imaging
E′ septal, cm/sec	11.0 ± 2.4	12.9 ± 11.8	NS
S′ septal, cm/sec	8.6 ± 1.5	8.2 ± 2.2	NS
E/E′ septal	8.1 ± 3.1	7.6 ± 2.1	NS

Abbreviations: A, late transmitral flow velocity; DT, deceleration time; E, early transmitral flow velocity; E′, early diastolic tissue Doppler velocity of mitral annulus; IVS, interventricular septum; LA, left atrium; LVEDd, left ventricular end‐diastolic diameter; LVEF, left ventricular ejection fraction; LVESd, left ventricular end‐systolic diameter; LVOT VTI, left ventricular outflow tract velocity time integrals; NS, not significant; PW, posterior wall; RWT, relative wall thickness; S′, systolic tissue Doppler peak velocity of mitral septal annulus; SOB, shortness of breath.

Figure 1.

Individual data on 3 parameters that were significantly different between healthy pregnant women with and without SOB (all P < 0.01). Abbreviations: CONT, control group; SOB, shortness of breath.

Similar normal longitudinal 2DS and 2DSR were found in both groups, as well as circumferential strain (Table 3). However, increased rotation at base level (P = 0.04) and a trend toward greater torsion (P = 0.09) were obtained.

Table 3.

Comparison of 2D Strain Measures Between Pregnant Women With SOB and Controls

	SOB Group, n = 30	Control Group, n = 30	P Value
Longitudinal
LGS, %	19.0 ± 3.6	19.6 ± 3.3	NS
LGSRs 1/sec	−1.1 ± 0.2	−1.1 ± 0.3	NS
LGSRe 1/sec	1.5 ± 0.4	1.5 ± 0.3	NS
LGSRa 1/sec	0.8 ± 0.2	0.8 ± 0.3	NS
Circumferential
CircGS, % (apex)	−21 ± 5.1	−20.1 ± 4.5	NS
CircGSRs 1/sec	−1.4 ± 0.4	−1.4 ± 0.3	NS
CircGSRe 1/sec	0.7 ± 0.4	0.6 ± 0.3	NS
CircGSRa 1/sec	0.7 ± 0.4	0.6 ± 0.3	NS
Rotation (apex), (°)	3.2 ± 5.6	4.0 ± 5.0	NS
CircGS, % PM	−16.3 ± 3.2	−15.3 ± 3.7	NS
CircGSRs 1/sec	−1 ± 0.2	−1.1 ± 0.3	NS
CircGSRe 1/sec	1.2 ± 0.4	1.1 ± 0.5	NS
CircGSRa 1/sec	0.5 ± 0.3	0.5 ± 0.3	NS
Rotation (PM), (°)	0.1 ± 2.7	1.7 ± 3.8	0.07
CircGS, % (base)	−15.5 ± 3.3	−13.9 ± 3.7	NS
CircGSRs 1/sec	−1 ± 0.3	−1 ± 0.3	NS
CircGSRe 1/sec	1.3 ± 0.4	1.1 ± 0.4	NS
CircGSRa 1/sec	0.5 ± 0.3	0.4 ± 0.3	NS
Rotation (base) (°)	3.5 ± 3.5	1.4 ± 3.9	0.04
Torsion (°)	9.3 ± 7.5	5.3 ± 6.5	0.09

Abbreviations: CircGS, circumferential global strain; CircGSR, circumferential global strain rate (a, late diastolic; e, early diastolic; s, systolic); GS, global strain; LGS, longitudinal global strain; LGSR, longitudinal global strain rate (a, late diastolic; e, early diastolic; s, systolic); NS, not significant; PM, papillary muscle; SOB, shortness of breath; SR, strain rate.

Discussion

Shortness of breath during pregnancy is very common and very bothersome to pregnant women, yet no single etiology for it has been determined, and thus no effective therapy has been devised. In this study we attempted to determine if subclinical myocardial dysfunction may be present and cause SOB in otherwise healthy pregnant women. We found that, compared with pregnant women who did not complain of SOB, those with significant SOB had thicker hearts and elevated pulmonary artery pressures, a significantly lower E‐wave DT and increased rotation with a trend toward a greater torsion. These may hint at the presence of subtle diastolic dysfunction that could lead to severe SOB and intolerance to stress during pregnancy.

Initially, SOB of pregnancy was attributed to displacement of the diaphragm and compression of the lungs by the uterus.5, 6, 7 However, because SOB may occur at any stage of pregnancy, not necessarily in the third trimester, other factors were sought.4, 5 A change in perception of normal respiration and hyperventilation secondary to reduced diffusion capacity were suggested6, 7 but has not been confirmed. Later, a higher sensitivity of central chemoreflex responses to CO₂ was found to contribute to dyspnea during pregnancy.8, 9 In addition, a good correlation between SOB and both progesterone and estradiol levels has been reported, suggesting that gestational hormones were involved in modulating the drive to breath.8 All the above‐mentioned studies focused on respiratory or central causes for SOB and did not consider a possible role of subtle cardiac factors. We found no extracardiac differences between the groups of pregnant women with and without SOB that could explain their SOB. Obstetrical‐related factors (maternal weight, fetal weight and birth weight, fetal Apgar score) and all other parameters studied were not different between groups and were well within normal limits.

A number of hemodynamic changes occur during pregnancy: blood volume and heart rate increase, and peripheral vascular resistance decreases.10, 11 These changes result in a significant increase in cardiac output, mostly due to the increase in HR. Diastolic function is also modified because of a rise in preload, decrease in afterload, and ventricular remodeling. To meet the increased demand on the heart and as part of LV remodeling, an increase in LV dimensions and mass has been reported, which may increase cardiac stiffness, impeding diastolic function. Changes in LV function in pregnant women with gestational hypertension and preeclampsia have been reported, indicating some alteration of LV geometry and diastolic function in this population.11 However, most of the data available on cardiac diastolic function during normal pregnancies12, 13 were derived from echocardiographic parameters that were not very sensitive to minor changes in LV diastolic function and were highly dependent on loading conditions.

During the last decade, TDI parameters of diastolic function have been shown to be relatively independent of changes in ventricular loading conditions and to have a good correlation with invasive hemodynamic measurements.14 Also, the recently developed 2DS technique based on speckle tracking provides information on multidimensional myocardial mechanics, including data on longitudinal and circumferential myocardial deformation as well as on rotation and torsion.15 Such indexes, known as deformation indexes, are now recognized to be very useful for assessment of early changes in LV function that cannot be detected by standard echocardiography. For instance, subclinical myocardial dysfunction based on impaired deformation indexes was reported in patients with nonalcoholic fatty liver disease with normal LVEF.16 Moreover, all studies that used 2DS to evaluate early changes in cardiac function in patients who received cardiotoxic chemotherapy uniformly demonstrate that changes in myocardial deformation indices precede changes in LVEF,17 and that a 10% to 15% early reduction in global strain during therapy appears to be the most useful parameter for prediction of cardiotoxicity and HF.

Recently, using this echocardiographic technique, a small decrease in LV segmental longitudinal systolic and diastolic function,18, 19 as well as changes in torsion and peak LV twist velocity, in hearts of women with otherwise normal pregnancies were reported.20, 21, 22 Thus, in the current study we employed indexes derived from both TDI and speckle tracking techniques, in addition to the more traditional methods, to study cardiac function in a unique group of patients to whom these methods have not yet been applied: those with significant SOB with no other known systemic or obstetric problem and no obvious reason on clinical investigation.

Cardiology today recognizes a continuum from mild HF due to diastolic dysfunction with preserved LV systolic function to overt systolic and diastolic dysfunction as in classical peripartum cardiomyopathy. The clinical symptom in common to all patients along this spectrum is dyspnea. For lack of sensitive tools for determining subtle cardiac dysfunction, other reasons for the sensation of dyspnea during pregnancy were sought over the years, and some were found. However, a cardiac source for either mild or severe SOB of pregnancy was never ruled out as a possible etiology, once peripartum cardiomyopathy, pulmonary embolism, or other overt cardiac problems were ruled out—it simply could not be ruled in. In this study, we noted differences between pregnant women with clinically significant SOB that in most cases required hospital admission and a matched control group without SOB, yet the absolute values of both groups were within the normal range, possibly because of the few hours', delay between admission and echocardiographic study. Thus, we may have missed the maximally abnormal values and obtained measurements after partial normalization of their cardiac dysfunction. Our findings of a greater wall thickness, decreased E‐wave DT, higher pulmonary artery pressure, an increase rotation of the base, and a trend for greater torsion may point to subtle myocardial systolic and diastolic dysfunction, perhaps a result of cardiac maladaptation to the increased hemodynamic burden imposed on the heart during pregnancy. This borderline LV dysfunction, although compensated at rest, may deteriorate on excursion and become clinically expressed as SOB during pregnancy.

Global longitudinal and circumferential strain did not differ significantly in our group of patients with SOB from the control group; however, increased rotation of the base and a trend for greater torsion in the former group is in accord with reports on similar findings in patients with diastolic dysfunction.21, 23 Fonseca et al24 found greater LV torsion by magnetic resonance imaging in patients with type 2 diabetes mellitus with diastolic dysfunction and normal LVEF than in a control group. Tzemos et al25 reported increased twist during pregnancy in women with bicuspid and stenotic aortic valves. Thus, the increased rotation with a trend to greater torsion in the SOB group compared with the controls may reflect a compensatory mechanism for borderline HF.

In contrast to the other etiologies proposed for SOB of pregnancy, if it is secondary to HF, then although not curable, it is treatable, hence the importance of proving or disproving our finding that myocardial dysfunction is involved in SOB of pregnancy.

Study Limitations

First, we had no way to objectively quantify SOB. However, the dyspnea must have been very significant to bring these pregnant women to the ED and to impress the ED doctors that pulmonary embolism or cardiomyopathy must be ruled out as a source of the SOB. Also, the short exercise duration of their stress tests is objective evidence of their exercise intolerance. For lack of an objective way to quantify the SOB, we could not estimate the correlation between the degree of SOB and echocardiographic findings. Second, this study was not designed as a longitudinal study, and so there are no follow‐up data on these patients and we have no data on progression or regression of their SOB after the index visit to the ED—although all were well enough to be discharged from the hospital once an acute cardiac or pulmonary reason was ruled out. Third, although statistically significantly different from the control group, none of the echocardiographic findings in the SOB group were outside the “normal” range. However, “normal” values by gestational age for each echocardiographic parameter are lacking, so we do not really know if they are truly normal.

Conclusion

In this study we report a trend suggesting subclinical changes in LV structure and subtle systolic and diastolic dysfunction in some women with significant SOB of pregnancy. Although these changes are subclinical at rest, they may become clinically relevant when there is an increased demand on the heart. These changes may result from cardiac maladaptation to the physiological hemodynamic changes of normal pregnancy. This hypothesis needs to be verified in a larger study.