Interictal dysphoric disorder and periictal dysphoric symptoms in patients with epilepsy

Authors


Address correspondence to Dr. Marco Mula MD, PhD, Department of Neurology, Amedeo Avogadro University, C.so Mazzini, 18 – 28100 Novara, Italy. E-mail: marco.mula@med.unipmn.it

Summary

Purpose:  The issue of phenomenology of mood disorders in epilepsy still remains controversial. It has been suggested that a subgroup of patients may develop an affective syndrome also known as interictal dysphoric disorder (IDD). However, the number of behavioral changes that may occur around the ictus needs to be taken into account for an accurate distinction between “true” psychiatric phenomenology and periictal phenomena. This study aimed at identifying clinical correlates of the IDD, with special attention to the relationship between symptoms and seizures.

Methods:  A sample of 142 consecutive adult outpatients with epilepsy were assessed using the Interictal Dysphoric Disorder Inventory (IDDI), a 38-item, self-report questionnaire specifically developed to evaluate presence and severity of IDD symptoms as well as their habitual association with seizures (coded as before, after, during, or when seizure-free) and their duration.

Results:  IDD was diagnosed in 31 subjects but symptoms showed a clear-cut relationship with epileptic seizures in 54.8% of cases, leading to an operative distinction between true IDD and periictal dysphoric symptoms (PDS). There was no significant difference among patients with IDD, PDS, or those without psychopathology. In the IDD group, symptoms were chronic and unremitting in one-third of cases, with labile affective symptoms being correlated with age at onset of seizures (rho = −0.612, p = 0.020) and duration of the epilepsy (rho = 0.833, p < 0.001).

Discussion:  An operative distinction between IDD and PDS bears the opportunity to identify different clinical endophenotypes that may have different prognoses and require different treatment strategies.

The issue of phenomenology of mood disorders in epilepsy still remains a matter of debate, with a number of authors pointing out that, especially among patients with refractory epilepsy, depression frequently fails to meet widely standardized criteria for psychiatric disorders such as those presented in the DSM (Mendez et al., 1986; Kanner et al., 2000; Kanner, 2003; Mula et al., 2008a). However, a number of studies have shown that a diagnosis of mood disorder could be allocated applying DSM criteria in a not negligible proportion of patients (Edeh & Toone, 1987; Victoroff et al., 1994; Jones et al., 2005).

In general terms, it is reasonable to hypothesize that patients with epilepsy can experience forms of mood disorders identical to those of patients without epilepsy. Nevertheless, it is equally reasonable to assume that the underlying brain pathology can influence the expression of psychiatric symptoms, making less evident some aspects or emphasizing others.

It has been suggested that, within the overall presentations of depression in epilepsy, a subgroup of patients may develop an affective syndrome characterized by dysphoric symptoms (Blumer, 2000). Kraepelin (1923), and then Bleuler (1949), were the first authors to describe a pleomorphic pattern of symptoms, including affective symptoms with prominent irritability intermixed with euphoric mood, fear, anxiety, as well as anergia, pain, and insomnia in patients with epilepsy. Gastaut et al. (1955) confirmed the observations of Kraepelin and Bleuler, leading Blumer to coin the term interictal dysphoric disorder (IDD) to refer to this type of affective syndrome (Blumer, 2000).

IDD is featured by eight key symptoms grouped into three major categories: labile depressive symptoms (depressed mood, anergia, pain, insomnia), labile affective symptoms (panic-like symptoms and anxiety), and supposedly “specific” symptoms (paroxysmal irritability and instable-euphoric moods). In a recent study, we observed that, although not specific for epilepsy itself, IDD seems to be a robust psychopathologic construct affecting a significant proportion of patients with epilepsy, thus being a frequent psychiatric comorbidity (Mula et al., 2008a). However, the number of behavioral changes that may occur around the ictus need to be taken into account (Trimble, 2004; Kanner, 2008) because the lack of an accurate distinction between “true” psychiatric phenomenology and periictal phenomena may lead to misinterpretations of symptoms (Mula et al., 2008b). This issue may have not only theoretical relevance but also relevant implications in terms of treatment strategies and prognosis.

Therefore, in this study, we sought to investigate clinical correlates of IDD in patients with epilepsy, with special attention to the relationship between IDD symptoms and seizures, in order to make an operative distinction between true IDD and periictal dysphoric symptoms (PDS). In addition, the role of the age of onset of seizures and a diagnosis of temporal lobe epilepsy has been investigated to identify a peculiar clinical phenotype.

Methods

Study sample

Data were drawn from a multicenter, cross-sectional study performed in adult outpatients with epilepsy, aimed at investigating clinical and psychopathologic features of affective disorders in epilepsy (Mula et al., 2008a). Patients were recruited in two tertiary referral centers in Europe, namely the Amedeo Avogadro University Department of Neurology in Novara, Italy and the Universitätsmedizin-Charité Department of Neurology in Berlin, Germany. To be enrolled, patients had to fulfill the following criteria: (1) diagnosis of epilepsy according to the International League Against Epilepsy (ILAE) criteria (1989); (2) age >18 years; (3) no gross cognitive deficits and a reading ability sufficient to manage the questionnaire, defined as a reading level greater than the 6th grade and/or a MMSE (Mini Mental State Examination score) >24; (4) absence of severe medical diseases; (5) lack of a high risk of suicide; and (6) willingness to provide written informed consent to undergo the experimental procedures.

Assessment procedures

All subjects were assessed with the Interictal Dysphoric Disorder Inventory (IDDI) and the Beck Depression Inventory (BDI) (Beck et al., 1961). The BDI was used to rate severity of depressive symptoms. The IDDI is a 38-item, self-report questionnaire specifically developed to diagnose IDD and to evaluate IDD symptoms in terms of presence, frequency, severity, and global impairment (Mula et al., 2008a). This questionnaire explores a time-interval of 12 months. The diagnosis of IDD follows Blumer’s criteria (Blumer et al., 2004), namely, the presence of at least three symptoms of “moderate” or “severe” severity and causing “moderate” or “severe” distress. It is possible to obtain a total score and three subscale scores that mirror the three major symptom categories described by Blumer: labile depressive symptoms, labile affective symptoms, and specific symptoms. Furthermore, the IDDI allocates also IDDI “severeness” (IDDIsev) total and partial scores that reflect the degree of interference or distress caused by IDD symptoms. The IDDI total and subscale scores showed strong or very strong correlations among themselves (0.68–0.85). The instrument showed an acceptable sensitivity and an excellent specificity when compared to validated questionnaire for the screening of major depression or bipolar disorder (i.e., BDI and Mood Disorder Questionnaire) (Mula et al., 2008a).

Finally, in the Appendix to the questionnaire, six questions investigate the time course of the disorder, the duration of dysphoric symptoms, and their associations with seizures or AED therapy. In particular, patients were asked whether dysphoric symptoms were habitually related only to seizures; if yes, in what temporal relation they occurred (coded as before, after, during, or when seizure-free), and how long dysphoric symptoms last (coded as a few hours, 1 day, a few days but less than a week, a week or more, chronic, variable duration). There was no specific definition for seizure freedom.

The entire questionnaire has been published and is fully available in Mula & Trimble, 2008.

The assessment was conducted by research clinicians who were trained in the use of the study instrument under the supervision of a neuropsychiatrist trained in neurology and psychiatry (MM, BS).

Statistical analyses

IDDI total and subscale scores were compared across epilepsy diagnoses looking specifically at differences among temporal lobe epilepsy (TLE), focal epilepsies other than TLE, and idiopathic generalized epilepsy (IGE) (all defined according to ILAE criteria).

Relationships between IDD symptoms and seizures were noted to identify different groups of subjects, namely true IDD and PDS. Patients with PDS have been operatively defined by the occurrence of at least three symptoms of at least “moderate” or “severe” severity and causing “moderate” or “severe” distress (i.e., Blumer’s criteria for IDD) referred by the subject as habitually related only to seizures.

The two identified groups were compared for gender, age of onset of seizures, epilepsy diagnoses, antiepileptic drug (AED) treatment with compounds recognized as mood stabilizers (carbamazepine CBZ, oxcarbazepine OXC, lamotrigine LTG, valproate VPA) versus no mood stabilizers (other AEDs), electroencephalography (EEG) interictal abnormalities in terms of localization and laterality, magnetic resonance imaging (MRI) features (coded as normal, right hemisphere lesion, left hemisphere lesion, bilateral lesions, unspecific lesions), seizure frequency (coded as free; 1–10 year; 1–10 month; 11–20 month; >20 month).

Frequencies of categorical demographic and clinical variables were analyzed using the chi-square analysis or Fisher’s exact test. Continuous demographic and clinical variables, IDDI total, and IDDI subscale scores were compared using the Mann-Whitney test or the Kruskal-Wallis test. Correlations between age, age at onset of the epilepsy, and IDDI total and partial scores were tested using the bivariate two-tailed nonparametric correlation procedures (Spearman’s coefficient). The alpha error was set at 0.05. All statistical analyses were two-sided and conducted using the Statistical Package for Social Sciences (SPSS) (version 12 for Windows, SPSS Inc., Chicago, IL, U.S.A.).

Results

We analyzed data of 142 patients (60 M, 82 F), with a mean age [± standard deviation (SD)] of 42.6 years (±14.4), and a mean age at onset of epilepsy of 23.0 years (±18.1); 102 were diagnosed partial epilepsy of whom 65 had temporal lobe epilepsy (Table 1). There was no between-group difference in the two centers (Berlin and Novara) in terms of age, gender, duration of the disease, marital status, education level, employment, distribution of epilepsy diagnoses, seizure type, and frequency.

Table 1.   Demographic and clinical characteristics of the study sample
 Total sampleIDDPDS
n = 142 (%)n = 14 (%)n = 17 (%)
  1. CBZ, carbamazepine; EEG, electroencephalogram; IDD, interictal dysphoric disorder; LTG, lamotrigine; MRI, magnetic resonance imaging; OXCBZ, oxcarbazepine; PDS, periictal dysphoric symptoms; TLE, temporal lobe epilepsy; VPA, valproate. All comparisons between IDD and PDS were not statistically significant. IDD, true interictal dysphoric disorder.

Gender
 Male60 (42.3)7 (50)8 (47.1)
 Female82 (57.7)7 (50)9 (52.9)
Age, years ± SD42.6 ± 14.444.4 ± 12.743.7 ± 9.5
Age onset of epilepsy, years ± (SD)23.0 ± 18.122.6 ± 20.717.9 ± 13.7
Epilepsy syndrome
 Cryptogenic partial66 (46.4)7 (50.3)9 (52.9)
 Symptomatic partial37 (25.7)4 (28.4)5 (29.4)
 Idiopathic generalized31 (22.1)2 (14.2)1 (5.9)
 Symptomatic generalized6 (4.3)1 (7.1)1 (5.9)
 Undetermined2 (1.4)01 (5.9)
Diagnosis of TLE65 (45.8)6 (42.9)11 (64.7)
Seizure frequency
 Free22 (15.4)3 (21.4)2 (11.8)
 <10/year46 (32.4)3 (21.4)6 (35.3)
 1–10/month62 (43.4)6 (43)8 (47)
 11–20/month6 (4.4)1 (7.1)1 (5.9)
 >20/month6 (4.4)1 (7.1)0
MRI features
 Normal107 (75.4)10 (71.5)12 (70.6)
 Right hemisphere lesion12 (8.2)1 (7.1)3 (17.6)
 Left hemisphere lesion17 (12.3)3 (21.4)1 (5.9)
 Bilateral lesions1 (0.9)00
 Unspecific lesions (e.g., distributed vascular abnormalities)5 (3.2)01 (5.9)
EEG
 Normal49 (34.8)6 (43.2)6 (35.3)
 Right24 (16.9)2 (14.2)3 (17.6)
 Left29 (20.1)2 (14.2)3 (17.6)
 Bilateral22 (15.3)2 (14.2)4 (23.6)
 Generalized18 (12.9)2 (14.2)1 (5.9)
AED therapy
 No therapy3 (2.1)1 (7.1)0
 Monotherapy92 (64.8)7 (50)11 (64.7)
 Dual therapy34 (23.9)4 (28.6)5 (29.4)
 Polytherapy13 (9.2)2 (14.3)1 (5.9)
 Mood stabilizing AEDs (LTG, CBZ, OXCBZ, VPA)121 (85.2)12 (85.7)12 (70.6)

The majority of patients (41.5%) were taking CBZ, and 28.2% VPA, 19% LTG, 16.2% levetiracetam (LEV), 11.3% topiramate, 9.9% barbiturate, and 7.7% OXC. In general terms, 85.2% of patients were taking at least one AED with recognized mood-stabilizing properties (CBZ, VPA, LTG, or OXC) (Table 1).

In the whole study sample, 54 patients (38%) were currently depressed, with a mean BDI score ± SD of 18.8 ± 6.6.

IDD symptoms were highly present in the whole population, with 83.8% reporting at least one labile depressive symptom, 60.6% at least one labile affective symptom, and 59.2% at least one specific symptom (Fig. 1). There was no difference in relationship with gender or AED therapy (mood stabilizers vs. not mood stabilizers); however, a clear-cut relationship with seizures was referred by more than one-half of patients in each symptom group (Fig. 1). Distribution of IDD symptoms in relationship to seizures is shown in Table 2.

Figure 1.


Prevalence of interictal dysphoric disorder (IDD) symptoms and their relationship to seizures in the whole sample (in darker gray, seizure-unrelated symptoms).

Table 2.   Periictal dysphoric symptoms in the study sample
IDD symptomsRelationship with seizuresn (%)
  1. IDD, interictal dysphoric disorder.

Labile depressive symptoms (depressed mood, anergia, pain, insomnia)Before19 (13.3)
After17 (12)
During11 (7.7)
When free7 (4.9)
Labile affective symptoms (panic-like symptoms and anxiety)Before16 (11.2)
After12 (8.4)
During9 (6.3)
When free5 (3.5)
Specific symptoms (paroxysmal irritability and instable-euphoric moods)Before19 (13.3)
After8 (5.6)
During10 (7)
When free7 (4.9 )

Patients with TLE showed higher IDDI total and subscale scores as compared to subjects with focal epilepsies other than TLE (F-noTLE) and IGE (IDDI tot = TLE 4.2 ± 2.2, F-noTLE 3.4 ± 2.0, IGE 2.6 ± 2.2, d.f. = 2, chi-square = 10.194, p = 0.006; Labile depressive symptoms = TLE 2.2 ± 1.2, F-noTLE 2.1 ± 1.3, IGE 1.4 ± 1.3, d.f. = 2, chi-square = 8.330, p = 0.016; Labile affective symptoms = TLE 1.0 ± 0.8, F-noTLE 0.8 ± 0.7, IGE 0.5 ± 0.6, d.f. = 2, chi-square = 10.390, p = 0.006; Specific symptoms = TLE 0.9 ± 0.7, F-noTLE 0.5 ± 0.6, IGE 0.7 ± 0.7, d.f. = 2, chi-square = 7.611, p = 0.022) (Fig. 2A). However, excluding from the analysis those subjects whose symptoms showed a clear-cut relationship with seizures, all differences among the three groups were no longer evident (IDDI tot = TLE 3.7 ± 2.1, F-noTLE 3.5 ± 1.7, IGE 2.7 ± 2.4, d.f. = 2, chi-square = 2.289, p = 0.318; Labile depressive symptoms = TLE 2.1 ± 1.2, F-noTLE 2.2 ± 1.2, IGE 2.7 ± 2.4, d.f. = 2, chi-square = 2.556, p = 0.279; Labile affective symptoms = TLE 0.9 ± 0.8, F-noTLE 0.9 ± 0.8, IGE 0.6 ± 0.8, d.f. = 2, chi-square = 2.462, p = 0.292; Specific symptoms = TLE 0.7 ± 0.7, F-noTLE 0.4 ± 0.5, IGE 0.6 ± 0.7, d.f. = 2, chi-square = 1.172, p = 0.557) (Fig. 2B).

Figure 2.


Interictal Dysphoric Disorder Inventory (IDDI) total and subscale scores in patients with temporal lobe epilepsy (TLE), focal epilepsies other than TLE (F-noTLE) and idiopathic generalized epilepsy (IGE). (A) In the total study sample. (B) After excluding patients whose symptoms showed a clear-cut relationship with seizures.

As for a formal diagnosis of IDD, 31 patients (21.8%) fulfilled Blumer’s criteria (i.e., the presence of at least three symptoms of at least “moderate” or “severe” severity and causing “moderate” or “severe” distress). However, in 17 patients, symptoms were reported as habitually related only to epileptic seizures, in particular 22.6% as preceding, 16.1% as following, 6.5% when seizure-free, and 3.2% during seizures (Fig. 3). Two patients reported a clear-cut relationship with their seizures but could not further specify whether before, after, or during seizures. Therefore, we distinguished two groups of patients, namely 14 with a true IDD and 17 with PDS, with a prevalence of 9.8% for the former and 12% for the latter. In the IDD group, symptoms were described as short-lasting (i.e., from a few hours to a day) in 21.4%, as chronic and unremitting in 35.7%, and as highly variable in duration in 42.9%. Conversely, the majority of patients with PDS (58.8%) described their symptoms as short-lasting (i.e., from a few hours to a day), whereas only 11.8% described their symptoms as chronic. There was no difference in severity of symptoms between IDD and PDS as assessed with the IDDIsev total and subscale scores. There was no difference in clinical or demographic characteristics in patients with IDD as compared to those with PDS or those without. In particular, there was no difference in age at onset of seizures, epilepsy syndrome diagnoses, MRI features, seizure type, seizure frequency, AED regimen, and combinations (see Table 1). However, in the IDD group, IDDI labile affective symptom scores strongly correlated with age at onset of seizures (rho = −0.612, p = 0.020) and duration of the epilepsy (rho = 0.833, p < 0.001). No significant correlations were found in the PDS group (age at onset of epilepsy rho = −0.098, p = 0.709; duration of epilepsy rho = −0.195, p = 0.452).

Figure 3.


Relationship between affective symptoms and seizures in patients with interictal dysphoric syndrome (IDD) as defined by Blumer’s criteria. NOS, not otherwise specified.

Discussion

To the best of our knowledge, this is the first study investigating in detail the clinical correlates of the so-called interictal dysphoric disorder (IDD) in a consecutive sample of adult patients with epilepsy.

The relationship between affective symptoms and seizures is a relevant issue addressed by our study. It is well known that a number of behavioral changes may occur around the ictus (Trimble, 2004; Mula & Monaco, 2007; Kanner, 2008), and several authors have attempted to identify and describe such phenomena (Blanchet & Frommer, 1986; Kanner et al., 2004; Mula et al., 2006). A cross-sectional study of patients in a monitoring unit showed a 43% prevalence of a mean of five symptoms of depression in the postictal period (Kanner et al., 2004). Moreover, the same authors reported also postictal manic symptoms in a noticeable proportion of subjects. Preictal mood changes are also well-known by clinicians and patients but, on the contrary, the prevalence is unknown, with the last study published about 20 years ago (Blanchet & Frommer, 1986). As shown in Table 2, IDD symptoms preceding seizures are not rare and, in some cases, are even more common than postictal ones, supporting the need of further studies investigating this pattern of symptoms preceding seizures. Regarding subjects in the seizure-free group, we are aware of the limits in our ability to infer forced normalization. However, patients were asked first whether their symptoms were habitually related only to seizures; if they answered yes, they were subsequently asked in what temporal relation these symptoms occurred. In our opinion, it is plausible to hypothesize that subjects who identified their symptoms as habitually related to seizure control are a different entity compared to those subjects who presented interictal dysphoric symptoms.

Our data are of relevance in terms of psychiatric assessment of patients with epilepsy, pointing out that a careful evaluation of the relationship between mood symptoms and seizures is always needed in epilepsy. In a previous study, we suggested that the diagnosis of bipolar disorder may be overestimated if periictal mood changes are not taken into account (Mula et al., 2008b). In fact, of the 11.8% of DSM-based diagnosis of bipolar disorder, only 1.4% could be considered as a “pure” psychiatric disorder because in a large proportion of cases manic/hypomanic symptoms were temporally related to seizures (i.e., postictal manic or hypomanic states and preictal dysphoria).

Likewise, in this study, we observed that among those who satisfied Blumer’s criteria for IDD, about one-half of patients presented symptoms with a clear-cut relationship with epileptic seizures (Fig. 3). In our opinion, a diagnosis of IDD, in the strict sense, should be limited only to those subjects for whom psychiatric manifestations are independent of the wide range of behavioral changes that may occur around the epileptic attack and, for this reason, the distinction between IDD and PDS was postulated. Such a distinction was not pointed out or directly addressed in the published studies of Blumer. Both entities, namely IDD and PDS, are worth noting because of their prevalence (9.8% for IDD and 12% for PDS) and their similarity in the reported grade of severity and impairment.

Our results have implications not only for diagnosis but, above all, for prognosis and development of treatment strategies. In fact, although there does not seem to be a difference in the phenomenology of dysphoric symptoms between IDD and PDS, it is evident that seizure control is determinant for the management of affective manifestations related to seizures (i.e., PDS), whereas specific psychiatric treatments (e.g., psychopharmacologic or psychotherapeutic interventions) are needed for IDD, especially because symptoms are chronic and unremitting in one-third of cases. In the present study, the distinction between periictal symptoms and major depression has not been addressed, but it is clearly evident that further research is needed to elucidate such a differential diagnosis. One of the most frequent methodologic errors in studies of depression in epilepsy is the sole reliance on screening instruments in the diagnosis of depressive disorders. In fact, a positive screening on the BDI may be the result of different situations such as side effects of antiepileptic medications, a depressive episode in major depression, dysthymia, double depression, or bipolar disorder (Barry et al. 2008). We further support the need for a careful assessment of periictal symptoms because they are frequently reported by patients and may be sufficiently severe to reach clinical attention although PDS cannot be considered true psychopathology.

The distribution of IDDI scores in our population (see Fig. 2A,B) supports the view that a clear-cut distinction between “pure” psychiatric symptoms and seizure-related behavioral manifestations is needed. In fact, patients with TLE showed significantly higher IDDI total and subscale scores as compared to patients with focal epilepsies other than TLE (mainly frontal lobe epilepsy) and patients with IGE (Fig. 3A). However, this difference was not noticeable when patients who presented symptoms with a clear-cut relationship with seizures were excluded from the analyses (Fig. 3B). These findings contribute to the never-ending debate on the existence of epilepsy-specific psychiatric syndromes (from the Geschwind syndrome to the IDD). In our opinion, in line with that of other authors (Swinkels et al., 2006; Devinsky, 2007), the involvement of specific brain areas in epilepsy, such as the temporal lobes, is likely to play a role in the flavor of behavioral manifestations, intended as some peculiar features or collateral symptoms, rather than in the development of a well-defined psychiatric syndrome. Our study suggests that such collateral symptoms may sometimes be seizure-based behavioral manifestations that are likely to be present in TLE.

In this light, the strong correlation between age at onset of seizures or duration of the disease and interictal labile affective symptoms (i.e., anxiety, panic, avoidance behavior) is of particular interest and may have different explanations. On one hand, it is plausible to speculate that a long history of an early onset brain disorder affecting, directly or indirectly, networks that are also involved in the processing of emotions, could have played a specific role in the pathogenesis of these symptoms. On the other hand, it is also possible that the early onset of a still highly stigmatized brain disorder such as epilepsy may affect self-esteem and contribute to the development of dependent behavior and anxiety symptoms.

Our results need to be considered keeping in the context of the following limitations. First, the small sample size reduces the statistical power of our study and some differences may be more evident in large populations. Second, our findings may not be representative of epilepsy patients in general because our population represents a highly select sample from tertiary referral centers. Third, the identification of a relationship between affective symptoms and seizures relied on subjective impression of patients. Retrospective self-report is notoriously unreliable and can be significantly affected by, among other things, current mood state. However, some authors observed that the endorsement of lifetime symptoms of mood (both depression and mania) is stable over time and is not sensitive to changes in clinical status (Cassano et al., 2009). No data are available about the IDDI. Further studies are warranted to replicate our findings.

In conclusion, we pointed out that in a remarkable proportion of patients IDD symptoms seem to be part of the wide range of affective manifestations that may occur around the ictus, leading us to introduce the concept of periictal dysphoric symptoms (PDS). The true IDD seems to have a chronic course with labile affective symptoms being correlated with the age of onset and the duration of the epilepsy. Longitudinal studies are warranted to estimate the burden of such clinical entities (i.e., PDS and IDD) in terms of psychosocial and health-related quality-of-life measures in order to develop specific treatment strategies. Finally, the relationship between major depression or subsyndromal mood spectrum dimensions and IDD needs to be further clarified.

Acknowledgments

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. The authors are grateful to the International League Against Epilepsy.

Disclosure

There are no financial interests or any commercial association that might pose a conflict of interest between the authors and this manuscript.

Ancillary