Aim: Patients with chronic obstructive pulmonary disease (COPD) exhibit depressed heart rate variability (HRV) compared to controls. Comorbidities of cardiac origin may even worsen the diagnosis and prognosis in these patients. In this study we investigated whether HRV markers, in addition to clinical markers of disease severity and respiratory function, may account for the presence of cardiac-related comorbidities. Methods: Temporal and spectral HRV indices were evaluated in 46 COPD patients (11 with comorbidities) before starting a six-minute walk test (6MWT). The maximum heart rate (HRmax) and the patient walking distance were also measured at the end of the test, while the heart rate recovery (HRR) index was estimated at specific time instants during recuperation. These parameters together with clinical markers of respiratory function and patient characteristics were used to discriminate patients presenting cardiac-related comorbidities (COPDco). A logistic regression classifier was used with relevant features selected via the Lasso regularization approach, while the model assessment was performed through a leave-one-out cross-validation (LOOCV). Results: A sparse model of only 4 features was able to discriminate the presence of comorbidities with overall performance metrics AUC=84%, sensitivity=77% specificity=83%. Features ordered by predictive power were: (1) the ratio given by the forced expiratory volume (FEV1) and the forced vital capacity (FVC): 45.2±12.9% (COPDco) vs 61.2±11.6% (COPDnco), p=0.001; (2) the HRR evaluated at minute 3 normalized by HRmax (23.7±5.9 vs 16.8±7.4%, p=0.005); (3) the Borg-scale of exertional dyspnea (2.4±1.7 vs 4.3±2.2, p=0.007), and (4) the normalized power in the LF band from HRV analysis (46.5±26.7% vs 58.3±22.5%, p=0.151). Conclusions: Cardiac comorbidities in COPD patients seem to be associated with clearly differentiated measures of lung function capacity, heart-rate recovery during recuperation, perception of exertion, and the sympathetic tone of autonomic control assessed during baseline. These features could provide relevant information for early identification of cardiac comorbidities.