Bacille Calmette Guerin (BCG) and prevention of types 1 and 2 diabetes: Results of two observational studies

Abstract

Background

Diabetes is a common disease marked by high blood sugars. An earlier clinical trial in type 1 diabetic subjects (T1Ds) found that repeat BCG vaccinations succeeded in lowering HbA1c values over a multi-year course. Here we seek to determine whether BCG therapy for bladder cancer may improve blood sugar levels in patients with comorbid T1D and type 2 diabetes (T2D). We also investigate whether BCG exposure may reduce onset of T1D and T2D by examining country-by-country impact of BCG childhood vaccination policies in relation to disease incidence.

Methods and findings

We first analyzed three large US patient datasets (Optum Labs data [N = 45 million], Massachusetts General Brigham [N = 6.5 million], and Quest Diagnostics [N = 263 million adults]), by sorting out subjects with documented T1D (N = 19) or T2D (N = 106) undergoing BCG therapy for bladder cancer, and then by retrospectively assessing BCG’s subsequent year-by-year impact on blood sugar trends. Additionally, we performed an ecological analysis of global data to assess the country-by-country associations between mandatory neonatal BCG vaccination programs and T1D and T2D incidence. Multi-dose BCG therapy in adults with comorbid diabetes and bladder cancer was associated with multi-year and stable lowering of HbA1c in T1Ds, but not in T2Ds. The lack of a similar benefit in T2D may be due to concurrent administration of the diabetes drug metformin, which inhibits BCG’s beneficial effect on glycolysis pathways. Countries with mandatory neonatal BCG vaccination policies had a lower incidence of T1D in two international databases and a lower incidence of T2D in one of the databases.

Conclusions

The epidemiological evidence analyzed here suggests that BCG may play a role in the prevention of T1D. It does not support prevention of T2D, most likely because of interference by metformin. Our ecological analysis of global data suggests a role for neonatal BCG in the prevention of T1D and, to a lesser extent, T2D. Randomized clinical trials are needed to confirm these findings.

Introduction

In 2018, approximately 10.5% of the US population had diabetes (around 34.2 million people). Of that total number, almost 1.6 million had Type 1 Diabetes (T1D), previously known as insulin-dependent diabetes or juvenile diabetes. There were about 1.5 million new cases of Type 2 Diabetes (T2D) in 2018, and about 18,000 new cases of T1D in children and adolescents aged below 20 years in 2014–2015 [1]. T1D is characterized by insulin deficiency due to an autoimmune response against the insulin-producing β islet cells in the pancreas, resulting in the inhibition of glucose uptake from the blood into cells. T2D early in the disease course is associated with insulin resistance; if poorly managed the disease can evolve to include insulin deficiency from β cells death, exhaustion, from non-immune causes [2]. Both types of diabetes, despite different etiology, are associated with insulin therapy and both types are associated with long term morbidity and mortality.

First used in 1921, the Bacillus-Calmette-Guerin (BCG) vaccine was designed to prevent tuberculosis (TB). It is a live attenuated vaccine developed from Mycobacterium bovis, an organism related to M. tuberculosis [3], which causes TB. The 100-year old BCG vaccine is the most commonly used vaccine in the world and enjoys a strong safety record. High-dose BCG has also been used for early stage bladder cancer therapy since the 1970s [4] with confirmed oncological benefits [5–7]. BCG vaccination has been associated with a range of off-target human clinical benefits, including improved protection from non-mycobacterial infections [8–10], reduced mortality in childhood, and increased protective effects for neonatal low birth weight boys [11], as well as clinical benefits in multiple sclerosis [12,13], T1D [14–17], and maybe even in Alzheimer’s disease [18,19].

The BCG vaccine has shown clinical benefits in T1Ds; one study associated multiple vaccinations during childhood with lower rates of T1D incidence [20], and within our group, two intradermal doses of BCG in advanced T1D patients corrected hemoglobin A1c (HbA1c) levels to near normal beginning after 3 years from treatment, with the positive effects persisting for a further 5 years [14]. Subsequent publications provide support for the reduction in blood sugar levels being due to both immune and metabolic effects. The immune effects include the reset of the immune system, inducing levels of suppressive T-regulatory (Treg) cells, and depletion of autoreactive cytotoxic lymphocytes (CTLs) that attack ß islet cells of the pancreas [15,21,22]. The immune reset is driven by de-methylation of 6 signature Treg genes in CD4 T cells [14]. Even in advanced T1D, when the pancreas no longer has functional ß islets, BCG therapy lowers blood sugars by a systemic reset of sugar metabolism. The BCG organism gradually and systemically switches glucose metabolism in immune cells from oxidative phosphorylation (OXPHOS) to aerobic glycolysis due to increased expression of Myc, a master regulator of several glucose metabolism pathways [16]. Notably, T1D patients at baseline show an increased state of OXPHOS compared to controls, a state of low glucose utilization. In contrast, the BCG-induced switch to aerobic glycolysis is a state of regulated high glucose utilization, promoting the anabolism of purines via the pentose phosphate shunt, which may consequently be driving down HbA1c levels. Interestingly, treatment with BCG did not induce hypoglycemia, a situational risk that may occur with increased standard insulin use for controlling blood sugar levels [14–17].

The reason for glucose underutilization in T1Ds could result from a lack of host-microbe interactions, a theory supported by the hygiene hypothesis [23]. It is only within the modern era that humans have been without frequent exposure to organisms commonly found in their environment. More specifically, the co-evolution between humans and Mycobacterium dates back around 90,000 years [24]. The data suggests that this interaction shapes the host immune system through a process of immune and metabolism reset. In many ways the reset of the immune system with the BCG vaccine mimics the localized effects of tuberculosis that also switches monocytes to aerobic glycolysis and turns on Treg cells [14,16]. The clinical permanence of BCG’s systemic effects may be dependent on epigenetic changes that are beneficial for both the host and the microbe [25]. In further support for this hypothesis, populations with a high prevalence of M. tuberculosis are better protected against T1D and other autoimmune diseases [26,27], and a stark difference in T1D incidence is evidenced due to changes in environmental exposure between the recently separated populations of eastern Finland and Russia [28].

In this study, we seek further support of BCG’s role in prevention of T1D. We also study BCG’s benefits for prevention of T2D. First, we studied three health care databases to look for the impact on blood sugars in elderly patients receiving high-dose BCG therapy for bladder cancer therapy in the US. Then on the global level we investigate the possible relatedness of neonatal BCG vaccines with subsequent diabetes prevention by tracking country-by-country incidence of T1D and T2D.

Methods

BCG-treated bladder cancer patients with existing T1D or T2D

Bladder cancer in the United States at an early clinical stage is commonly treated with a six-course regimen of large doses of BCG applied weekly into the bladder. The average age for the diagnosis of bladder cancer is 73, and it is more common in men than women [29]. To explore the hypothesis that high-dose intravesical BCG in bladder cancer subjects might alter existing blood sugar control in either T1D or T2D, we analyzed three large clinical care datasets in the US. The datasets consist of the Research Patient Data Registry (RPDR) from the Massachusetts General Brigham (MGB) system, from Management Science Associates (with Quest Diagnostics data) and data from Optum Labs (OL). The RPDR is a centralized clinical data registry that gathers data from hospital systems across the MGB network and stores it in one place, enabling researchers to pull a variety of patient data (from 1988 onwards), including and not limited to, patient demographics, diagnoses, medications and procedures. Quest Diagnostics and Optum Labs data also have their own private datasets that helped to investigate the hypothesis of this study. All these datasets were from predominantly US based and US born patients. This was important to prevent the inclusion of a patient population that had received BCG neonatal vaccinations due to health care policies. The US has never had a policy of mandatory neonatal BCG vaccination.

Since massive datasets obtained from clinical care may contain errors in coding between T1D and T2D we set up multiple layers of data searching to ensure the subjects were properly categorized. We made sure the T1Ds were never on oral diabetes drugs and always used insulin. The MGB system is the largest healthcare provider in Massachusetts, USA, with treatment of over a third of the population within the Boston metropolitan area. The data registry was searched for patients who were diagnosed with bladder cancer, with a comorbid T1D or T2D diagnosis, who were then split into two groups consisting of those who had BCG treatment for bladder cancer and those who did not (Table 1). This data is also represented in S1 Fig. For the purposes of this study any dose of BCG when used for bladder cancer qualified the patient into the BCG-treated group. Along with the search criteria for patients, HbA1c results were requested from the databases for analysis and then analyzed for evidence of prevention. For the data to be of use, multi-year, bi-annual and annual HbA1c values had to also be available. The use and study of patient data from the RPDR was approved by the MGB Institutional Review Board (number: 2001P001379).

Table 1. Search criteria for data from various data sources.

		T1D Diagnosis	T2D Diagnosis	Insulin Only	T2D Medication	Bladder Cancer	BCG for Bladder Cancer	BCG for Other Purposes
MSA	T1D	+		+	-		+	-
	T2D		+		+		+	-
MGB	T1D	+	-	+	-	+	+	-
	T2D	-	+	-	+	+	+	-
OL	T1D	+			-	+	+

Bladder cancer and BCG treatment data were also obtained independently from two other US-based health care sources, OL and MSA (with Quest Diagnostics). Optum Labs data includes around 45 million individuals. Quest Diagnostics is derived from the world’s largest database of clinical lab results, with access to around 90% of insured individuals in the US (N = 263 million) and more than 6600 patient access points (predominantly in the US, but also with operations in the UK, Brazil and Mexico). The different databases used in this study were analyzed and filtered independently from each other using varying inclusion and exclusion criteria (Table 1).

Roughly 76,000 T1D patients were identified in the RPDR database. The search for T1D with bladder cancer, involved having a T1D diagnosis, as well as being only on insulin and no other oral diabetic drugs (Table 1). BCG was also added as an inclusion criterion when used for bladder cancer but subjects were excluded when BCG was used for another purpose. HbA1c results were requested for the patients identified. Additionally, there were around 367,000 T2D patients in the same database. These patients were filtered to identify those who were not on any insulin, and who received BCG for bladder cancer treatment and not for any other purpose. The T1D patients were all male and had an average age of 78 at the time of receiving BCG treatment. The patients identified with T2D were 78% male and 21% female; the average age at the time of BCG administration was 71.9 years, 71.2 years for males and 74.8 years for females. HbA1c data for the RPDR dataset ranged from 1998 to 2020.

The search using the Optum Labs data identified patients between 2016 and April 2021. Patients were searched for using a T1D diagnosis (ICD10 E10.9), a bladder cancer diagnosis (c679) and also the code for intravesical BCG instillation (J9031 and J9030). As this treatment is only used for patients with bladder cancer, it is safe to assume these patients were diagnosed with the disease. The patients were then further selected based on the number of T1D diagnoses they have had since 2016, compared to the number of T2D diagnoses, those with a greater ratio of type 1 were selected for further analysis. Additionally, patients were also not on any oral medications for diabetes.

The MSA data were gathered from T1D patients in the US and between the ages of 40–82 between 2014 and 2020. Patients were required to receive BCG for bladder cancer treatment and no other purpose. T1Ds were precluded from receiving any oral diabetes medication and were only on insulin. The search identified roughly 374,386 T1D patients, and down to a further 145,226 patients who had their HbA1cs with Quest Diagnostics. Of the T1D patients analyzed, their average age was 71 years. In contrast, T2D had a T2D diagnosis, and were on insulin and at least one oral medication for diabetes. In the majority of cases this oral drug was metformin. There were 2,214,116 T2D patients when unrestricted with the insulin filter; when filtering with insulin and a T2D medication, a total of 541,343 T2D patients were identified. The average age for the T2D patients was also 71 years old.

Data from the RPDR, OL and MSA were anonymized. Analysis of the data from the RPDR was approved by our Autoimmune and Metabolic Disorders: In Vitro Pathogenesis and Early Detection protocol number: 2001P001379.

BCG neonatal vaccination policies and T1D and T2D incidence, a global analysis

We asked if neonatal vaccines were mandatory for a country, and what was the incidence of T1D or T2D in that country. This was an ecological study looking at country-level associations without adjustment for potentially confounding factors. Data for the global incidence of T1D was gathered from diabetes.org.uk (which features data from the International Diabetes Federation (IDF)) [30], where the incidence of T1D and T2D per 100,000 children ages 0 to 14 is listed by country. The BCG status of each country was established using the world BCG Atlas website [31], and independent internet searches. The incidence information gathered for T1Ds was repeated in the Global Health Data Exchange (GHDx) [32] for the same countries (barring Hong Kong, a province of China). Additionally, based on the same countries, T2D incidence data for people of all ages was also downloaded and analyzed. The GHDx database also had a larger profile of countries to study, and so T1D (ages 0–14) and T2D (all ages) incidence data for 204 countries and territories was assessed. Although these datasets may not be independent from each other, they provide a perspective of the incidence at two different time points; 2011 for the IDF dataset and 2019 for the GHDx dataset.

Statistical analysis

For the bladder cancer patient databases, each individual dataset was analyzed separately and then combined. A Student’s paired t-test was used to compare average HbA1cs pre-BCG treatment to average HbA1cs each subsequent year afterward. All error bars represent the standard error of the mean. To determine the impact of BCG as a neonatal vaccine and diabetes prevention, the data between countries that currently administer BCG and those that do not were compared using a Mann-Whitney U test. All data processing was performed in Microsoft Excel version 16.43. All statistical analysis and graphing were finalized in Prism version 9.1.1.

Results

BCG treatment for bladder cancer in T1D is associated with lower blood sugar

Fig 1 presents results from the three large clinical databases of subjects with either T1D (top) or T2D (bottom) who also received BCG for bladder cancer (Figs 1 and 2). Bladder cancer occurs in the elderly [29], so this was a population of diabetic subjects with long standing disease. The blood sugar control post-multi-dosing BCG for bladder cancer, was monitored as a percent change in HbA1c value. The HbA1c for each subsequent year of a patient is compared to their pre-BCG baseline.

Fig 1. Lower HbA1cs are observed in Type 1, but not Type 2, diabetic subjects post-BCG treatment for bladder cancer.

For T1Ds, all three datasets (MSA, RPDR and OL) show a reduction in HbA1c, calculated as percentage change in HbA1c values post-BCG instillation. Each dataset also shows a near 10% decrease in HbA1cs at differing time points. Combined, the data for T1Ds convey a statistically significant decrease in year 1 post-BCG instillation (p = 0.0304). In contrast, the MSA data for T2Ds show no change in the HbA1cs of patients, and the RPDR reveals an increase in HbA1c values post-BCG instillation. The combined data shows no significant change. N’s for each dataset: RPDR = 4, MSA = 9 and OL = 6 for T1Ds. And RPDR = 97, MSA = 9 for T2Ds.

Fig 2. Decreasing trends in HbA1c are visible when using the average HbA1c for each dataset for T1Ds, but not T2Ds.

In T1Ds, lower HbA1cs are seen for all three databases for subjects treated with BCG for bladder cancer. T2Ds on the other hand, do not show any trends post BCG instillation. N’s for each dataset; RPDR = 4, MSA = 9 and OL = 6 for T1Ds. And RPDR = 97, MSA = 9 for T2Ds.

All three datasets, MSA, MBG and OL, show a near 10% decrease in HbA1c value after treatment for bladder cancer with the BCG instillation. The MSA patient data is observed to have a near 10% decrease in HbA1c within the first year itself, and persists with a decreased HbA1c for the following two years. And within the MGB database, patient data appeared to show a gradual decrease in HbA1c over time, not including the data point for year 3.

The data from the MBG database appears to gradually decrease HbA1c at a higher percentage each year after, aside from year three. A 10% decrease in HbA1c appears to be reached by year 6 after BCG treatment. Similarly, the Optum Labs data also shows a gradual decrease in HbA1c, reaching a near 10% decrease by year 3. Naturally, for those years that are available in all three datasets, combining the searched T1D data of subjects receiving high dose BCG for bladder cancer once again showed from three separate databases, a combined graph where every year it is observed that after high dose BCG for bladder cancer there is a decrease in the HbA1c value compared to a pre-BCG baseline. With regards to the combined data for T1Ds (n = 19), this decrease is significant in year 1 (p = 0.0304).

In contrast, no decrease in HbA1c values was seen in T2Ds after BCG treatment for bladder cancer. T2Ds for the MSA database presented no change in year one, followed with an increase in HbA1c in year two. Year three for the T2D MSA data showed a decrease, however, the error bar is quite large. Additionally, the T2D RPDR data appeared to increase HbA1c values post BCG instillation. An increase over 5% was observed in years 4 and 6. As expected, combining the data for the two T2Ds subject sets (n = 106) from two separate health care databases, using yearly data available in both sets, displayed no change in HbA1c for T2Ds.

The average HbA1c values for each year post-BCG are graphed in Fig 2. For T1Ds, the graphs present lowered values compared to year 0. For T2Ds, the average HbA1cs exhibit far more sporadic changes.

Lower incidence of T1D in countries with neonatal BCG vaccination and mixed results for T2D in an ecological analysis of the global population

For the ecological study of T1D, we looked at two datasets involving the incidence of T1D in children aged between 0 to 14 years within various countries from two separate databases. The countries were divided into those with a policy of mandatory neonatal BCG vaccination and those without. The search was on the IDF database and matching countries and then repeated in the GHDx, and then further expanded to all GHDx countries and territories (Fig 3). T1D incidence in countries currently administering BCG and those that are not were compared. The data show the incidence of T1D around the world on a country-by-country basis appears to be associated with current BCG vaccination programs at birth. The incidence rate of T1D per country is graphed out in Fig 3. The graphs are then split into those countries that currently have a BCG vaccination program and those that do not. Countries that currently administer BCG have a significantly lower T1D incidence rate than those that do not administer the vaccine at birth. Data from the IDF database showed an average of 65% reduced incidence for T1D in countries with a BCG vaccination policy (p<0.0001). When matched to the GHDx database, countries with a BCG vaccination policy had an average of 39% reduced incidence for T1D (p<0.0001), and when further compared with all the GHDx countries and territories, those with a BCG vaccination policy had an average of 47% reduced incidence for the disease (p<0.0001). Although the data presented in this article do not allow a conclusion that BCG has a preventing effect on T1D, the epidemiological data suggests lowered T1D incidence in countries with neonatal BCG vaccine programs and provokes further investigation.

Fig 3. In global population data BCG neonatal vaccinations consistently correlate with reduced incidence of T1D.

From the IDF list of countries, those with current BCG vaccination programs (blue lines) are observed to have significantly reduced incidence of T1D in children (p = <0.0001). Countries with a mandatory BCG vaccination policy had an average of 65% reduced incidence. An almost identical trend was observed with a repeat search of T1D incidence for children in the GHDx database, using the same countries as the IDF database. This lends support to prevention based on the result that countries with a childhood BCG program are observed to have an average of 39% lower incidence for T1D (p = <0.0001). Finally using all the countries and territories available in the GHDx dataset, we again observe the incidence of T1D in children is reduced by an average of 47% if they earlier received neonatal vaccinations of BCG (p = <0.0001). A Mann-Whitney U test was used to compare the two groups of countries with (blue) and without (black) newborn BCG vaccinations based on the country-by-country policies. Black lines represent countries without BCG neonatal vaccination programs; blue lines represent countries with BCG vaccination programs. [N = 33 for BCG- countries, N = 56 for BCG+ for the IDF database] [N = 45 for BCG-, N = 159 for BCG+ countries and territories in the GHDx database]. Notation of the countries are in Table 2.

The same analysis was also conducted for T2D (Fig 4). The T2D incidence was extracted for individuals of all ages for the same countries. Furthermore, the searches were conducted on the IDF-matched countries, and all GHDx countries and territories. Fig 4 depicts the incidence of T2D for all ages in the same countries as Fig 3. The IDF-matched countries database did not observe a difference in T2D incidence between neonatal vaccinated and non-vaccinated countries (p = 0.0715). When this comparison was repeated for all the countries available in the GHDx database, the p-value was significant (<0.0001), indicating a reduced incidence rate for countries with mandatory neonatal BCG vaccination, with an average reduction of 28% for those countries with a BCG vaccination policy. Taken together, this suggests a possible prevention benefit for T2D. Keep in mind for this data, the BCG vaccine was administered at birth with no concurrent metformin administrations. The order of countries for Figs 3 and 4 is listed in Tables 2 and 3.

Fig 4. In global population data BCG neonatal vaccinations might confer some protection from T2D onset.

Using IDF-matched countries in the GHDx database, the T2D incidence was not significantly different between countries with (in blue) a childhood BCG program and those without (in black) neonatal BCG vaccinations (p = 0.0715). However, using all the countries and territories available in the GHDx dataset, the T2D incidence was reduced by an average of 28% in those countries with neonatal vaccine programs (p <0.0001). A Mann-Whitney U test was used to compare the two groups of countries with and without newborn BCG vaccinations policies. Black lines represent countries without BCG neonatal vaccination programs; blue lines represent countries with BCG vaccination programs. [N = 33 for BCG- countries, N = 55 for BCG+ for the IDF database] [N = 45 for BCG-, N = 159 for BCG+ countries and territories in the GHDx database]. Notation of the countries are in Table 3.

Table 2. Countries detailed in Fig 3.

Fig 3A			Fig 3B			Fig 3C			Fig 3C
Order	Country	T1D Incidence (per 100k) 2011	Order	Country	T1D Incidence (per 100k) 2019	Order	Country	T1D Incidence (per 100k) 2019	Order	Country	T1D Incidence (per 100k) 2019
1	Barbados	2	1	Barbados	8.41849118	1	Northern Mariana Islands	4.18835184	103	Saint Vincent and the Grenadines	8.67328198
2	Antigua and Barbuda	3.5	2	Antigua and Barbuda	8.70963679	2	Tokelau	4.20796262	104	Somalia	8.68626835
3	Switzerland	9.2	3	Bahamas	9.27104451	3	Guam	4.31530128	105	Guinea-Bissau	8.68680832
4	Bahamas	10.1	4	United States Virgin Islands	9.93560871	4	American Samoa	4.35048748	106	Haiti	8.70623606
5	Israel	10.4	5	Puerto Rico	10.6648363	5	Barbados	8.41849118	107	Burundi	8.71874258
6	Greece	10.4	6	Slovakia	15.0130732	6	Suriname	8.62936004	108	United Republic of Tanzania	8.72005937
7	Slovenia	11.1	7	Slovenia	15.0751843	7	Grenada	8.67911654	109	Cabo Verde	8.72130717
8	Italy	12.1	8	Israel	15.8554815	8	Antigua and Barbuda	8.70963679	110	Uganda	8.72181381
9	France	12.2	9	Czechia	16.0478769	9	Trinidad and Tobago	8.93334296	111	Djibouti	8.76534908
10	United States Virgin Islands	12.8	10	New Zealand	25.3153067	10	Bahamas	9.27104451	112	Gambia	8.7776097
11	Spain	13	11	Germany	25.6531714	11	United States Virgin Islands	9.93560871	113	Saint Kitts and Nevis	8.78717951
12	Portugal	13.2	12	Iceland	25.7251128	12	Puerto Rico	10.6648363	114	Saint Lucia	8.83072477
13	Austria	13.3	13	Belgium	25.9164017	13	Slovakia	15.0130731	115	Togo	8.86263591
14	Slovakia	13.6	14	France	25.921243	14	Slovenia	15.0751843	116	Jamaica	8.91437359
15	Iceland	14.7	15	Austria	25.9629784	15	Israel	15.8554815	117	Cameroon	8.94333177
16	Cyprus	14.9	16	Greece	26.3366991	16	Czechia	16.0478768	118	Madagascar	8.97033822
17	Belgium	15.4	17	Luxembourg	26.3532542	17	Lebanon	16.7415742	119	Eritrea	8.98483386
18	Luxembourg	15.5	18	Netherlands	26.3776591	18	Bahrain	17.9833834	120	Liberia	8.99835639
19	Malta	15.6	19	Cyprus	26.431014	19	New Zealand	25.3153067	121	Rwanda	9.0050404
20	Ireland	16.3	20	Denmark	26.5502525	20	Germany	25.6531714	122	Zambia	9.01369502
21	Puerto Rico	16.8	21	United Kingdom	26.7348637	21	Iceland	25.7251128	123	Malawi	9.03273765
22	Czech Republic	17.2	22	Sweden	27.2272774	22	Belgium	25.9164017	124	Mauritania	9.0776351
23	Germany	18	23	Switzerland	27.2884586	23	France	25.921243	125	Ghana	9.15442077
24	New Zealand	18	24	Portugal	27.5966609	24	Austria	25.9629784	126	Dominica	9.18477377
25	Netherlands	18.6	25	Malta	27.8392689	25	Greece	26.336699	127	Sao Tome and Principe	9.25060153
26	Canada	21.7	26	Australia	27.8942744	26	Luxembourg	26.3532542	128	Comoros	9.28775103
27	Denmark	22.2	27	Italy	28.0518214	27	Netherlands	26.3776591	129	Angola	9.29918429
28	Australia	22.5	28	Ireland	28.122439	28	Cyprus	26.431014	130	Ethiopia	9.33474185
29	USA	23.7	29	Norway	28.3166011	29	Denmark	26.5502525	131	Central African Republic	9.35748283
30	United Kingdom	24.5	30	Spain	28.6532107	30	United Kingdom	26.7348637	132	Bermuda	9.41813187
31	Norway	27.9	31	United States of America	29.0649277	31	Monaco	27.0358802	133	Namibia	9.59188688
32	Sweden	43.1	32	Canada	31.0183544	32	San Marino	27.1652697	134	Democratic Republic of the Congo	9.63402172
33	Finland	57.6	33	Finland	31.5968355	33	Sweden	27.2272774	135	Eswatini	9.65112868
34	Papua New Guinea	0.1	34	Papua New Guinea	3.98583625	34	Switzerland	27.2884586	136	Nigeria	9.65727353
35	Venezuala	0.1	35	Mauritius	4.12490665	35	Portugal	27.5966609	137	Botswana	9.66049643
36	Ethiopia	0.3	36	Thailand	4.146415	36	Malta	27.8392689	138	Gabon	9.74772167
37	Thailand	0.3	37	China	5.49599318	37	Australia	27.8942744	139	Zimbabwe	9.74948953
38	Dominican Republic	0.5	38	Taiwan (Province of China)	6.90977276	38	Italy	28.0518213	140	Kenya	9.8773046
39	Pakistan	0.5	39	Mexico	7.47271717	39	Ireland	28.122439	141	Lesotho	9.88576217
40	Peru	0.5	40	Venezuela (Bolivarian Republic of)	7.61608618	40	Norway	28.3166011	142	Equatorial Guinea	9.88752346
41	China	0.6	41	Colombia	7.74799261	41	Andorra	28.606577	143	Congo	10.0343312
42	Zambia	0.8	42	Dominican Republic	8.23238719	42	Spain	28.6532107	144	Bhutan	10.2945796
43	United Republic of Tanzania	0.9	43	Cuba	8.2405577	43	United States of America	29.0649277	145	Nepal	10.451862
44	Paraguay	0.9	44	Peru	8.25246473	44	Canada	31.0183544	146	Bangladesh	10.5849429
45	Republic of Korea	1.1	45	United Republic of Tanzania	8.72005937	45	Finland	31.5968355	147	South Africa	10.6596328
46	Uzbekistan	1.2	46	Zambia	9.01369502	46	Maldives	3.80229625	148	Pakistan	11.2080274
47	Tajikistan	1.2	47	Dominica	9.18477377	47	Papua New Guinea	3.98583625	149	India	11.6793415
48	Colombia	1.3	48	Ethiopia	9.33474185	48	Cambodia	3.98619758	150	Belarus	12.426027
49	Mauritius	1.4	49	Nigeria	9.65727353	49	Lao People’s Democratic Republic	4.01704167	151	Lithuania	12.5022363
50	Mexico	1.5	50	Pakistan	11.2080274	50	Viet Nam	4.03035699	152	Latvia	12.5031145
51	China, Hong Kong	2	51	India	11.6793415	51	Solomon Islands	4.05839697	153	Ukraine	12.6527634
52	Cuba	2.3	52	Belarus	12.426027	52	Myanmar	4.07567062	154	Republic of Moldova	12.922527
53	Japan	2.4	53	Lithuania	12.5022363	53	Timor-Leste	4.07873098	155	Mongolia	12.9493839
54	Oman	2.5	54	Latvia	12.5031145	54	Mauritius	4.12490665	156	Turkmenistan	12.9793999
55	Singapore	2.5	55	Ukraine	12.6527634	55	Malaysia	4.12673836	157	Tajikistan	13.1091225
56	Nigeria	2.9	56	Tajikistan	13.1091225	56	Thailand	4.146415	158	Uzbekistan	13.2220909
57	Jordan	3.2	57	Uzbekistan	13.2220909	57	Nauru	4.15246382	159	Paraguay	13.2587446
58	Bosnia and Herzegovina	3.5	58	Paraguay	13.2587446	58	Seychelles	4.17433692	160	Armenia	13.29554
59	Iran	3.7	59	Georgia	13.6724789	59	Sri Lanka	4.21320863	161	Kazakhstan	13.3061246
60	Taiwan (Province of China)	3.8	60	Estonia	13.8553586	60	Vanuatu	4.22973515	162	Kyrgyzstan (Kyrgyz Republic)	13.3178656
61	Macedonia	3.9	61	Brazil	14.8529997	61	Marshall Islands	4.25515197	163	Azerbaijan	13.344206
62	India	4.2	62	Romania	14.9076929	62	Philippines	4.28760962	164	Georgia	13.6724789
63	Georgia	4.6	63	North Macedonia	14.9698974	63	Tuvalu	4.30850765	165	Estonia	13.8553585
64	Romania	5.4	64	Russian Federation	15.1147867	64	Micronesia (Federated States of)	4.3315387	166	Albania	14.5077392
65	Belarus	5.6	65	Bosnia and Herzegovina	15.2086585	65	Fiji	4.35499885	167	Brazil	14.8529996
66	Dominica	5.7	66	Bulgaria	15.2647083	66	Kiribati	4.37833863	168	Romania	14.9076929
67	Chile	6.6	67	Poland	15.4046938	67	Cook Islands	4.37912312	169	North Macedonia	14.9698973
68	Argentina	6.8	68	Hungary	15.8071974	68	Tonga	4.39438024	170	Russian Federation	15.1147867
69	Tunisia	7.3	69	Montenegro	15.9314049	69	Samoa	4.40412569	171	Bosnia and Herzegovina	15.2086584
70	Latvia	7.5	70	Serbia	15.9747971	70	Indonesia	4.49515116	172	Bulgaria	15.2647083
71	Brazil	7.7	71	Croatia	16.0840552	71	Palau	4.52608185	173	Poland	15.4046938
72	Lithuania	7.8	72	Algeria	16.6917219	72	Niue	4.54221712	174	Hungary	15.8071974
73	Egypt	8	73	Oman	16.9535195	73	Democratic People’s Republic of Korea	5.47293252	175	Montenegro	15.9314049
74	Ukraine	8.1	74	Jordan	17.0755815	74	China	5.49599318	176	Serbia	15.9747971
75	Uruguay	8.3	75	Egypt	17.179117	75	Taiwan (Province of China)	6.90977276	177	Croatia	16.0840552
76	Algeria	8.6	76	Sudan	17.3656422	76	Mexico	7.47271717	178	Algeria	16.6917218
77	Libyan Arab Jamahiriya	9	77	Tunisia	17.5472166	77	Nicaragua	7.51598782	179	Oman	16.9535195
78	Croatia	9.1	78	Qatar	17.6847926	78	El Salvador	7.53964237	180	Jordan	17.0755814
79	Bulgaria	9.4	79	Iran	17.7977458	79	Costa Rica	7.5662181	181	Egypt	17.179117
80	Sudan	10.1	80	Libya	19.5986822	80	Venezuela (Bolivarian Republic of)	7.61608618	182	Iraq	17.1893454
81	Hungary	11.3	81	Kuwait	20.2935776	81	Panama	7.62589598	183	Palestine	17.2383524
82	Qatar	11.4	82	Saudi Arabia	22.1815331	82	Guatemala	7.64405328	184	Sudan	17.3656422
83	Russian Federation	12.1	83	Chile	22.640754	83	Honduras	7.65106803	185	Turkey	17.4584652
84	Serbia	12.9	84	Argentina	22.7204972	84	Colombia	7.74799261	186	Morocco	17.4729206
85	Montenegro	16.3	85	Japan	26.3844224	85	Niger	8.12097039	187	Yemen	17.4891509
86	Estonia	17.1	86	Uruguay	26.3960018	86	Mali	8.18155498	188	Afghanistan	17.5125341
87	Poland	17.3	87	Singapore	27.4131412	87	Bolivia (Plurinational State of)	8.19023698	189	Tunisia	17.5472166
88	Kuwait	22.3	88	Republic of Korea	27.7686293	88	Chad	8.22798653	190	Qatar	17.6847926
89	Saudi Arabia	31.4				89	Dominican Republic	8.23238719	191	Iran	17.7977458
						90	Cuba	8.24055769	192	Syrian Arab Republic	18.1549846
						91	Peru	8.25246473	193	Libya	19.5986821
						92	Burkina Faso	8.26404246	194	Kuwait	20.2935776
						93	Ecuador	8.29175985	195	United Arab Emirates	20.7737443
						94	Benin	8.34165073	196	Saudi Arabia	22.1815331
						95	Guinea	8.41347429	197	Chile	22.640754
						96	Sierra Leone	8.4954355	198	Argentina	22.7204972
						97	Mozambique	8.51199212	199	Japan	26.3844224
						98	Cote d’Ivoire	8.52158872	200	Uruguay	26.3960018
						99	Guyana	8.56839232	201	Greenland	26.8596142
						100	South Sudan	8.57490059	202	Brunei Darussalam	27.3010106
						101	Belize	8.6018442	203	Singapore	27.4131412
						102	Senegal	8.61772682	204	Republic of Korea	27.7686293

Table 3. Countries detailed in Fig 4.

Fig 4A			Fig 4b			Fig 4B
Order	Country	T2D Incidence (per 100k) 2019	Order	Country	T2D Incidence (per 100k) 2019	Order	Country	T2D Incidence (per 100k) 2019
1	France	198.565858	1	France	198.565858	103	Lesotho	229.625022
2	New Zealand	240.617148	2	New Zealand	240.617148	104	Tajikistan	230.399076
3	Australia	266.050564	3	Australia	266.050564	105	Maldives	231.546297
4	Netherlands	267.111533	4	Netherlands	267.111533	106	Bhutan	231.606347
5	Canada	270.897434	5	Canada	270.897434	107	Uruguay	232.950113
6	Denmark	273.753976	6	Denmark	273.753976	108	Estonia	235.654007
7	Israel	277.370361	7	Israel	277.370361	109	Eswatini	239.083084
8	Iceland	281.984628	8	Iceland	281.984628	110	Japan	241.671157
9	Ireland	288.16046	9	Ireland	288.16046	111	Democratic People’s Republic of Korea	242.219317
10	Sweden	304.116706	10	Sweden	304.116706	112	Latvia	242.345729
11	Belgium	310.210766	11	Belgium	310.210766	113	Indonesia	243.462195
12	Austria	320.212843	12	Austria	320.212843	114	Cabo Verde	243.69783
13	Slovakia	324.11394	13	Andorra	320.568032	115	Sudan	246.482366
14	Greece	324.265569	14	San Marino	322.157831	116	Afghanistan	252.186282
15	Switzerland	328.437205	15	Slovakia	324.11394	117	Egypt	254.070509
16	Norway	335.828799	16	Greece	324.265569	118	Uzbekistan	254.978501
17	Slovenia	342.37443	17	Switzerland	328.437205	119	Gabon	256.152213
18	Finland	400.99166	18	Monaco	330.01071	120	Paraguay	256.255215
19	Spain	425.708827	19	Norway	335.828799	121	Republic of Moldova	261.104295
20	Bahamas	443.975531	20	Slovenia	342.37443	122	Botswana	262.306924
21	Italy	449.412402	21	Guam	370.25036	123	Cambodia	262.539118
22	Cyprus	454.516272	22	Lebanon	390.888638	124	China	262.883254
23	United Kingdom	455.838349	23	Finland	400.99166	125	Greenland	270.932022
24	Malta	469.499462	24	Spain	425.708827	126	Lao People’s Democratic Republic	272.720823
25	Luxembourg	474.253636	25	Bahamas	443.975531	127	Romania	274.243091
26	United States of America	477.534067	26	Italy	449.412402	128	Ecuador	283.731311
27	Portugal	505.28596	27	Cyprus	454.516272	129	Argentina	289.204334
28	Germany	508.755645	28	United Kingdom	455.838349	130	Belize	289.628832
29	Antigua and Barbuda	530.630579	29	Malta	469.499462	131	Viet Nam	290.98144
30	Barbados	557.640251	30	Luxembourg	474.253636	132	Azerbaijan	298.596954
31	United States Virgin Islands	561.96423	31	United States of America	477.534067	133	India	302.477819
32	Czech Republic	633.503628	32	Tokelau	480.260351	134	Brazil	304.587822
33	Puerto Rico	634.398805	33	Portugal	505.28596	135	South Africa	306.132221
34	Ethiopia	79.9431839	34	Germany	508.755645	136	Honduras	308.914094
35	Nigeria	87.2144585	35	Antigua and Barbuda	530.630579	137	Kazakhstan	311.374169
36	United Republic of Tanzania	96.9322533	36	Suriname	545.544176	138	Palestine	313.077947
37	Zambia	112.478886	37	Barbados	557.640252	139	Turkey	317.4904
38	Belarus	169.493795	38	Grenada	559.205716	140	Oman	325.506074
39	Lithuania	189.2065	39	United States Virgin Islands	561.96423	141	Jordan	330.170255
40	Peru	190.41484	40	Northern Mariana Islands	611.248179	142	Myanmar	330.528975
41	Ukraine	190.475629	41	Czechia	633.503628	143	Armenia	330.709688
42	Russian Federation	191.262756	42	Puerto Rico	634.398805	144	Malaysia	332.823322
43	Dominican Republic	213.949069	43	Trinidad and Tobago	699.321142	145	Nicaragua	333.09696
44	Pakistan	220.89271	44	American Samoa	800.216785	146	Haiti	342.330931
45	Tajikistan	230.399076	45	Bahrain	996.44129	147	Guatemala	342.683139
46	Uruguay	232.950113	46	Niger	71.0183322	148	Iraq	344.31551
47	Estonia	235.654007	47	Ethiopia	79.9431839	149	Iran	345.776608
48	Japan	241.671157	48	Sierra Leone	86.1949884	150	El Salvador	350.893421
49	Latvia	242.345729	49	Nigeria	87.2144585	151	Syrian Arab Republic	356.364708
50	Sudan	246.482366	50	United Republic of Tanzania	96.9322533	152	Colombia	358.487405

Discussion

Two different types of epidemiologic datasets were used to test the hypothesis that BCG might have an impact on diabetes management or might prevent diabetes onset. These outcomes were measured through a drop in HbA1c values with existing diabetic disease or in a decrease in diabetes incidence. Within the US, BCG is only approved by the FDA for the treatment of bladder cancer. A multi-year observational study of HbA1c values in adult subjects shows post-BCG high dose bladder instillation improved blood sugar control in subjects with existing T1D, but not T2D. The bladder cancer subjects were elderly with longstanding diabetes yet still maintain BCG responsiveness in T1D. The multi-year time line of improved blood sugar mirrored multi-dose BCG as a vaccine in a double-blinded controlled trial (the average age of patients in the trial at the time of their baseline visit was 38 years old) [14]. BCG vaccines in neonates may also prevent T1D, according to our examination of global datasets. A country-by-country practice of BCG administered neonatally correlated with lower T1D incidence. The impact of neonatal BCG vaccines for prevention of T2D is more complicated.

The general mechanism behind improved HbA1c after BCG vaccinations in adult T1D may reflect, at least in longstanding disease, a systemic shift in metabolism. Juvenile-onset T1Ds have an underlying lymphoid defect in regulated sugar utilizations through a dependence on oxidative phosphorylation, a cellular energy step wherein cells use less sugar for energy [14,15]. With BCG administration this underlying defect in sugar utilization is reversed and the lymphoid system, both T cells and monocytes, shifts to nearly restored aerobic glycolysis [16]. The mechanism for improved HbA1c can be monitored by various complex or even simple methods such as sugar uptake in culture over set time periods with labeled sugar by registering fluorescence. For T2D, published data does not show an underlying defect in aerobic glycolysis suggesting a therapeutic effect of BCG may be less dramatic [33]. The capability of the lymphoid system being able to regulate blood glucose levels was hypothesized in 2007, where immune cells could transiently restrict the rise in blood glucose during and after a meal by buffering glucose in the form of lactate and aspartate [34,35]. This mechanism is different than the mechanism of converting oxidative phosphorylation to aerobic glycolysis, but illustrates the previous lack of appreciation of using the massive lymphoid organ system within humans as a new and novel regulatory system of blood sugars.

The BCG effects on blood sugars in T1D may involve the protein MYC, which stabilizes Hypoxia-inducible factor 1-alpha (HIF-1α), leading to increased uptake of glucose [16]. HIF-1α is activated by the mechanistic target of rapamycin (mTOR) and, typically, in response to food intake, the body will indirectly activate mTOR due to chain signaling from Insulin Receptor Substrate (IRS) to Phosphoinositide 3-kinase (PI3K) and then Protein Kinase B (Akt), resulting in the inactivation of Tuberous Sclerosis Complex 2 (TSC2) (which is a suppressor of mTOR), and thus activation of HIF-1α and glucose uptake. A recent publication reported that BCG affects the epigenetic methylation status of Lysine Demethylase 2B (KDM2B) to facilitate improved mTOR functionality [16]. BCG treatment, increased DNA methylation at a specific site on the KDM2B gene (cg13708645) back to levels comparable for non-diabetic controls, suggesting reduced expression of the KDM2B protein, leading to decreased demethylation of Histone 3 Lysine 3 (H3K4me3) and Histone 3 Lysine 36 (H3K36me2). These histone modifications result in the activation of cytokines, such as TNF, IL6 and TLR4, as well as increased glycolysis [36]. A specific CpG site for the gene DNA Damage Inducible Transcript 4 (DDIT4) (cg01674036) was also observed to have significant hyper-methylation towards non-diabetic control levels post BCG treatment. The consequence of this hyper-methylation suggests inhibited expression of the DDIT4 protein, a protein known to suppress mTOR activity. Activated mTOR is thus further able to boost the glucose uptake due to HIF-1α, and increase the rate of glycolysis. When comparing the two forms of diabetes, the methylation dysregulation for the CpG site in KDM2B is in the opposite direction in T2D [37] signifiying the differences between the two diseases, and may be a contributing factor to the increased OXPHOS. Interestingly, various obesity related factors have been shown to upregulate DDIT4, which may contribute towards the development of insulin resistance, and genetic inhibition of the DDIT4 gene has been observed to impair insulin sensitivity [38]. Re-methylation of the specific CpG site on DITT4 back to normal levels after BCG treatment, suggests a correction of the metabolism pathways in T1Ds.

There is also growing evidence that the immune system and inflammation are principal defects in Alzheimer’s and Parkinson’s disease [39,40]. Certainly neurons are heavily dependent on proper sugar uptake for survival. BCG, in both animal models and in epidemiology studies, suggests additional beneficial effects for adults. This could be one reason why BCG treatment for bladder cancer correlates with prevention of Alzheimer’s disease and multiple sclerosis, both conditions relating to the immune system [18,19,41].

Bladder cancer patients with existing T2D treated with high-dose BCG appear to show no change to their HbA1c after treatment. This result could be due to the biological differences between T1Ds and T2Ds, but there is another explanation that should be seriously considered. As mentioned above, BCG’s accelerated sugar utilization is controlled in part through the mTOR pathway [16]. Almost every T2D in the US takes metformin. Metformin interferes with this BCG-controlled metabolic pathway. Metformin appears to treat diabetes by reducing hepatic gluconeogenesis, although the full mechanism may be more complex [42]. Contrastingly, metformin also appears to inhibit the mTOR pathway by increasing expression of AMPK, an effect that would be inhibitory to the mechanism of BCG. Indeed Arts and colleagues show a detrimental effect of metformin when actually administering metformin to normal volunteers and then studying the in vitro effects on isolated monocytes [43]. Administering metformin, an AMPK (AMP-activated protein kinase) activator (and mTOR inhibitor), to healthy volunteers for 5 days yielded monocytes that no longer had the beneficial phenotype of BCG’s effect on innate immunity. Metformin inhibited the induction of altered immunity by BCG, as shown by a temporarily decreased induction of cytokine responses and lactate production upon secondary stimulation. We have also seen that metformin in vitro totally inhibits the normal BCG conversation of lymphoid cells from high oxidative phosphorylation to aerobic glycolysis [33]. This effect of metformin, now confirmed by two different research groups, may be the reason why reduced HbA1c in T2Ds are not observed post-BCG instillation for bladder cancer. But the protective trend can be observed in population based studies with neonatal vaccines, a time without metformin interference.

The molecular pathway for bladder cancer strongly appears to also involve expression of KDM2B, as well as the mTOR pathway [44,45]. And BCG’s further activation of the mTOR pathway appears contradictory towards the treatment of bladder cancer. However, the effect of BCG may be multi-faceted, either directly affecting the cancer cells, or epigenetically reprogramming peripheral blood cells to combat the tumorous cells more effectively. Interestingly, the mechanism of metformin appears advantageous for bladder cancer. It has been observed to potentially prolong the recurrence interval; however, the rate of recurrence itself did not appear affected [46]. With regards to diabetes, a recent Nature Vaccines publication did show beneficial effects of BCG on T2D mice, suggesting the need for further investigation [47]. Taken together, this data suggests that subjects in future double-blinded placebo-controlled trials studying the effects of BCG on T2D need to be free from metformin therapy.

The mechanism for BCG as a bladder therapy showing a beneficial effect described above may not fully reflect the story in bladder cancer, as multi-dosing may be a key aspect for the beneficial aspects around BCG. Even within our group, previous studies show that two-doses of BCG were needed to induce lowered HbA1c in humans [14]. This was also true many years ago even in mouse models of type 1 diabetes, multi-dosing led to better efficacy [48]. Patients undergoing treatment for non-muscle invasive bladder cancer have 6 high doses of BCG (50mg of BCG) administered over 6 weeks. A phase III clinical trial using reduced frequency of BCG instillations displayed increased likelihood for the recurrence of bladder cancer [49]. The data in the field suggests that multiple doses of BCG are needed for appropriate immune system re-programming.

An analysis of neonatal BCG vaccination programs and the incidence of T1D showed that countries currently administering the vaccine at birth had a lower incidence of this disease. The data are observational and, to our knowledge, no clinical trial of multi-dosing BCG in childhood to prevent T1D has been conducted. Previous studies in mice have also suggested the possibility that BCG may help prevent T1D [50–52]. Interestingly, children receiving at least two doses of the Moscow BCG strain, with the first dose being in the newborn period, have displayed a lower incidence of T1D in an observational study from Turkey [20]. Data from Greece also suggests childhood BCG administration may prevent T1D [53]. Therefore for the prevention of diabetes with BCG administered in childhood, the data supports multi-dosing.

BCG is a live attenuated vaccine and likely persists long term in the vaccinated host. The beneficial effects of the BCG vaccine might be emulated by other live vaccinations such as the Rotavirus vaccine [54]. The choice of the live vaccine for treatment of prevention of autoimmunity or other off-target beneficial effects of vaccines will likely relate to the long term safety of each vaccine, especially with regards to childhood vaccinations.

In conclusion, three different patient datasets showed decreased HbA1c’s post-BCG intravesical instillation for bladder cancer, even though the T1D subjects were elderly and had longstanding disease. The data suggests that BCG helps regulate blood sugars in T1D patients. We observed that multi-dosing of BCG elicited a beneficial response. Previous studies using a single dose of BCG did not show promising results, whereas those with multi-dosing have generated promising ones. The interval between doses, number of doses required, and BCG strain require further investigation. Additionally, more thorough investigations into the possibility of prevention of T1D and T2D should also be conducted based on the observed associations between neonatal BCG vaccinations programs and diabetes incidence. Because of the potential interference of metformin with the BCG mechanism of action, future randomized double blinded controlled trials should exclude metformin use by T2D subjects. Interestingly, a recent publication from Quebec, Canada [55], found that BCG vaccinated individuals were at a lower risk of T1D, and also might be at a lower risk for T2D as well. Further studies are warranted to validate the potential beneficial effects of BCG on diabetes, including the need for multiple doses, or a particular strain.

Limitations

All of the data assessed in this study were observational and causality cannot be implied. A number of potential confounders need to be addressed before making a conclusion and warrants further investigation into the observations we have come across in our study. Some of these confounders involve lifestyle choices, health expenditure, development, healthcare resource availability, diagnostic ability, and hunger index for T2Ds. Furthermore, since the majority of T2Ds are on metformin, we could not adequately assess BCG’s impact on prevention. Our analysis of the GHDx dataset suggests that the BCG’s promising results for prevention may translate to gains in prevention of T2D provided that patients are not taking metformin.

Supporting information

S1 Fig. Flowcharts of the selection criteria of patients.

Flowcharts showing the selection criteria set for each database to filter for either T1D or T2D patients.

(TIF)

Abbreviations

T1D

type 1 diabetes

T2D

type 2 diabetes

BCG

Bacillus Calmette-Guerin

CTLs

cytotoxic lymphocytes

MGB

Massachusetts General Brigham hospital system

GHDx

Global Health Data Exchange

IDF

International Diabetes Federation

US

United States

Data Availability

All data is in the paper.

Funding Statement

This study was funded by The Iacocca Foundation. Funding was awarded to DLF. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.