Cervical Cancer 2010–2019: Midwest Catchment of 40,000 square miles

Abstract

Objectives:

American Indian (AI) women have a higher incidence and mortality from cervical cancer than non-Hispanic White (NHW) women in the US. Our purpose is to detail atthe clinical events in the cervical cancer prevention continuum among the AI and White women with cervical cancer on the US frontier.

Methods:

A cancer center with a nearly 40,000 square-mile catchment area maintained a detailed cancer registry connected to the clinic records of all cervical cancer patients between 2010–2019. This catchment area provided records of both an AI and a White population. Descriptive and inferential statistics and modeling predictions detailed the prevention continuum.

Results:

Among the 126 with cervical cancer, 20% were AI, and 78% were White. 60% did not participate in cervical cancer screening within the five years before their diagnosis, and on average, 9.2 years passed since the last cervical cancer screening. 91% presented with symptoms, and most women presented with two or more symptoms. 13% underwent a colposcopic diagnostic step, significantly delaying the time to diagnosis compared to other diagnostic steps. 69% of the histopathologic diagnoses were squamous cell carcinoma, and 27% were adenocarcinoma. 49% presented at Stage I regardless of histopathology. Chemotherapy and radiation therapy were most commonly combined. 63% of the population survived, and 42% survived at least three years from diagnosis. Younger age and earlier Stages at diagnosis were the significant adjusted predictors of survival.

Conclusions:

Our detailed cervical cancer prevention continuum events provide new data questioning the use of colposcopy for women symptomatic at presentation.

Introduction

In the United States (US), American Indian (AI) people with a cervix (women) are 2.2% of the US population.¹ 87% live outside a tribal statistical area, and 13% are on reservations or trust lands. South Dakota, part of the Northern Plains tribal area, has the fourth-highest AI population at 9.8% ², serving nine reservations, with the three largest reservations on the state’s western side. There are nine non-NCI-designated cancer centers in South Dakota, and one is on the Western side.

AI women of the Northern Plains have more prevalent human papillomavirus (HPV) infection than the US general population, where 35% of the AI population is positive for at least one high-risk HPV type.³ South Dakota AI had a 42% HPV prevalent infection rate, and 32% of the HPV infections were not vaccine-preventable types, significantly higher than the non-Hispanic White (NHW) of the general US population.⁴

Over ten years of cytology-based screening (1999–2009), the purchased/referred care delivery area (PRCDAs) county data show a 1.5-fold increase in the incidence of cervical cancer among AI over NHW women (11/100,000 in AI vs. 7.1/100,000 in NHW) with variations among regions, For instance, AI in the Northern Plains have experienced the highest relative incidence of cervical cancer (1.97) compared to NHW.^5,6

From 2014–2018, the incidence worsened to 11.5 vs. 7.4/100,000, with AI women having a 56% higher rate of cervical cancer incidence in the PRCDA counties than NHW women of the general population.⁷ Specifically, AI women 35–49 years had a 1.5-fold increase in cervical cancer incidence over NHW women, and AI women 50–64 years had a 1.8-fold increase in incidence rate.⁸ AI disparities continue to exist, with Northern Plains regions having nearly twice the rate of cervical cancer incidence as NHW.⁹

By 2018, the incidence of cervical cancer among AI women had increased by 0.6%, an even higher rate than all cancers combined among the AI population.⁷ In addition, AI women have later-stage cervical cancer presentations than NHW women in the general US: a 1.8-fold increase at the regional Stage and a 2.4 increase in the distant Stage at diagnosis.^7,8 Between 2008 and 2017, using Urban Indian Health Organizations (UIHO), AI women living in UIHO service areas had a 1.5-fold increase in the risk of cervical cancer compared to NWH urban counterparts. Regional inequities continue to exist, with AI of the Northern Plains demonstrating an increased relative risk of 1.8 compared to NHW.¹⁰

The purpose of this study is to document the most recent decade of the steps in the cervical cancer prevention continuum in a rural population enriched with AI women.

Methods

The IRB approved the study as RH-23-92 in September 2023. We conducted a retrospective cross-sectional medical record review of all women in the tumor registry at the cancer center for cervical cancer treatment between January 1, 2010-December 31, 2019. We extracted the data from the electronic medical records as discrete data elements and scanned records from referring practices. Three extractors (MAP, PLW, LK) cross-reviewed 10% of the data for accuracy checks.

Data

Common data elements included age at diagnosis, self-identified race, tobacco use, health insurance payor at the time of diagnosis, and the distance from the cancer center. Ethnicity was not established. We separated the dates of cervical cancer screening into dates associated with the immediate cancer diagnosis and the next most recent screening date for those who participated in screening. We recorded all symptoms and their onset as recorded in the record. We recorded the colposcopy data, if done, or the date of blind biopsy or excisional treatment. We recorded the date of cancer diagnosis, Stage of cancer (International Federation of Obstetrics and Gynecology [FIGO] staged according to the definitions in place at the time of diagnosis), and type of cancer. We did not independently verify the Stage at which the diagnosis was recorded. Treatment data included the type of treatment, the number of types of treatment, the date of the first treatment, the date of treatment completion, and survival.

Statistical Analysis

Descriptive statistics included frequency distributions, summary statistics, and cross-tabulations for both continuous and categorical data. Inferential statistics included chi-square comparisons inferred with a two-sided alpha of 0.05. Survival predictive modeling used logistic regression. The study is not powered to see differences between AI and White women, but where significant differences were noted, they were reported. All statistical calculations were performed with Statistica v14.0.1.25¹¹.

Results

Of the 126 women identified with cervical cancer diagnoses between 2010 and 2019 from the Western South Dakota Cancer Center (Supplementary Figure 1), 25 (20%) self-identified as AI, 98 (77%) self-identified as White, and one each self-identified as Black, Asian, and Middle Eastern, making the ‘Other’ group of 3 (2%).

Table 1 presents the demographic data. Among the three populations, the same portion of women had never smoked (29%), had ever smoked (26%), and were currently smoking (38%). Supplementary Table 1 shows that smoking status was not associated with the Stage at diagnosis. White women had private insurance significantly more than AI women (47% vs. 8%, p<0.001), and AI women had public insurance significantly more than White women (88% vs 43%, p<0.001). Significantly more AI women lived in South Dakota than did White women (96% vs 64%, p<0.001). While there was no significant difference in distance between the woman’s residence and the treatment center, the only distance distribution significantly different was the mid-distance (50–99.9 miles) being more common among the AI than the White women (28% vs 7%, p=0.003). The mid-distance represents a measure of accessibility to cancer care for AI women.

Table 1.

Demographic descriptors of the population

	American Indian	White	Other (Black, Asian, Middle East)	Total
	N=25 Column %	N=98 Column %	N=3 Column %	N=126
Smoking Status N(%)				N=118
Never	7 (28%)	28 (29%)	2 (67%)	37 (29%)
Past	8 (32%)	25 (26%)	0 (0%)	33 (26%)
Current	10 (40%)	37 (38%)	1 (33%)	48 (38%)
Insurance* N(%)				N=126
Private	2 (8%)	46 (47%)	2 (67%)	50 (40%)
Public	22 (88%)	43 (43%)	1 (33%)	66 (52%)
None	1 (4%)	9 (10%)	0 (0%)	10 (8%)
State of Residence§ N(%)				N=125
South Dakota	24 (96%)	63 (64%)	1 (33%)	88 (71%)
North Dakota	0 (0%)	8 (8%)	2 (67%)	10 (8%)
Nebraska	1 (4%)	7 (7%)	0 (0%)	8 (6%)
Wyoming	0 (0%)	18 (18%)	0 (0%)	18 (14%)
Nevada	0 (0%)	1 (1%)	0 (0%)	1 (1%)
Distance from treatment center, miles (mean, SD)	55.3 (54.6)	124.4 (183.4)	184.3 (140.8)	112.0 (166.8)
Under 50 miles	12 (48%)	50 (52%)	1 (33%)	63 (50%)
50–99.9 milesǂ	7 (28%)	6 (6%)	0 (0%)	14 (11%)
100 or more miles	6 (24%)	41 (42%)	2 (67%)	49 (39%
Age at diagnosis (mean, SD)	49.9 (14.4)	54.4 (14.5)	45.8 (20.6)	53.3 (14.6)
Under 30 years	1 (4%)	5 (5%)	0 (0%)	6 (5%)
30 −39.9 years	7 (28%)	11 (11%)	2 (67%)	20 (16%)
40–49.9 years	7 (28%)	23 (23%)	0 (0%)	30 (24%)
50–59.9 years	3 (12%)	21 (21%)	0 (0%)	24 (19%)
60–65 year	3 (12%)	10 (10%)	0 (0%)	13 (10%)
Over 65 years	4 (16%)	28 (29%)	1 (33%)	33 (26%)
Range (min, max)	29, 82	23, 89	31, 69	23, 89

^*p<0.001 for AI is significantly different from White in private and public insurance

^§p<0.001 for AI is significantly different from White in South Dakota, state of residence

^ǂp=0.003 for AI are significantly more often in the mid-distance from the treatment center than White

The mean age at diagnosis was 53.3 years (SD14.6); the age distribution was similar among all populations. Over one-quarter of the cervical cancers diagnosed were in patients older than 65, and this was more frequent among NHW than AI (29% vs 16%). The most common age at diagnosis was 40–49.9 years among both AI and NHW, at 28% and 23%, respectively.

Table 2 displays the screening, workup, and symptom presentation for women with cervical cancer. Over half (52%) of women did not participate in screening before their diagnosis, and this did not vary by racial population. We do not have access to the screening results for those who did screen, but we report the type of diagnostic workup that followed. Of those with screening, 17% had no further workup (data not shown). Most women (79%) had a blind biopsy (random cervical biopsy without colposcopic guidance), 13% underwent a colposcopic exam guiding tissue excision, and 8% underwent excisional therapy.

Table 2.

Screening, work-up, and symptoms associated with cervical cancer presentation

	American Indian	White	Other (Black, Asian, Middle East)	Total
	N=25	N=98	N=3	N=126
Cervical Cancer Screening prior to diagnosis
Yes	12 (48%)	38 (39%)	0 (0%)	50 (40%)
No	13 (52%)	50 (51%)	2 (67%)	65 (51%)
Unknown	0 (0%)	10 (10%)	1 (33%)	11 (9%)
Diagnostic work-up after presentation N (%)	N=25	N=88	N=3	N=116
No work-up	5 (20%)	14 (16%)	1 (33%)	20 (17%)
Work-Up	20 (80%)	74 (84%)	2 (67&)	96 (83%)
Biopsy without colposcopy (includes EUA, EMB, ECC)*	17 (85%)	59 (80%)	0 (0%)	76 (79%)
Colposcopy+ (with biopsy or LEEP or cone)	3 (15%)	8 (11%)	1 (50%)	12 (13%)
Excisional Therapy (LEEP or cone)	0 (0%)	7 (9%)	1 (50%)	8 (8%)
Symptoms at diagnosis	N=52	N=167	N=6	N=225
Bleeding associated symptoms §	26 (50%)	90 (54%)	2 (33%)	118 (52%)
Intermenstrual or heavy vaginal bleeding	13 (50%)	55 (61%)	2 (100%)	70 (59%)
Post-coital bleeding	8 (31%)	24 (27%)	0 (0%)	32 (27%)
Post-menopausal bleeding	3 (6%)	10 (11%)	0 (0%)	13 (11%)
Rectal Bleeding	1 (2%)	1 (1%)	0 (0%)	2 (2%)
Anemia	1 (2%)	0 (0%)	0 (0%)	1 (1%)
Pain associated symptoms Ɨ	19 (37%)	46 (28%)	2 (33%)	67 (30%)
Pain- pelvic or abdominal	9 (47%)	22 (48%)	1 (50%)	32 (48%)
Urinary symptoms: back pain, UTI, hematuria	5 (26%)	15 (32%)	0 (0%)	20 (30%)
Dyspareunia	5 (26%)	9 (20%)	1 (50%)	15 (22%)
Constitutional symptoms ǂ	6 (12%)	18 (20%)	2 (33%)	26 (12%)
Weight loss/gain	3 (50%)	11 (60%)	1 (50%)	15 (57%)
Gastrointestinal symptoms: nausea/vomiting/diarrhea/constipation	0 (0%)	5 (28%)	0 (0%)	5 (19%)
Weakness/syncope/fatigue	2 (33%)	1 (6%)	1 (50%)	4 (15%)
Night Sweats	1 (17%)	1 (6%)	0 (0%)	2 (8%)
Other	1 (2%)	13 (14%)	0 (0%)	14 (6%)
Vaginal discharge	1 (100%)	12 (92%)	0 (0%)	13 (93%)
Deep venous thrombosis	0 (0%)	1 (8%)	0 (0%)	1 (7%)
Number of presenting symptoms per person	N=25	N=98	N=3	N=126
Zero	2 (8%)	8 (8%)	0 (0%)	10 (8%)
One	8 (32%)	37 (38%)	2 (67%)	47 (37%)
Two	6 (24%)	20 (20%)	0 (0%)	26 (21%)
Three	7 (28%)	17 (17%)	0 (0%)	24 (19%)
Four	0 (0%)	6 (6%)	1 (33%)	7 (6%)
Five	1 (4%)	3 (3%)	0 (0%)	4 (3%)
Six	1 (4%)	0 (0%)	0 (0%)	1 (1%)

^*EUA means exam/biopsy under anesthesia, EMB means endometrial biopsy, ECC means endocervical curettage

^§Bleeding symptoms presented significantly more often than pain (52% vs 30%, p<0.001) and intermenstrual or heavy vaginal bleeding presented more often than post-coital bleeding (59% vs 29%, p<0.001)

^ƗPelvic or abdominal pain presents more frequently than back pain and urinary symptoms (48% vs 30%, p<0.05)

^ǂWeight loss is significantly more common than other individual constitutional symptoms, p<0.01

Among all women, 225 symptoms were reported at diagnosis. All racial populations equally experienced all symptom types. All racial populations equally experienced all symptoms. 52% of the symptoms were associated with bleeding and presented as intermenstrual/heavy vaginal bleeding, post-coital bleeding, post-menopausal bleeding, rectal bleeding, and anemia. Intermenstrual or heavy vaginal bleeding presented more often than post-coital bleeding (59% vs 29%, p<0.001, Supplementary Table 2).

Bleeding symptoms presented significantly more often than pain (52% vs 30%, p<0.001). The most common sites of pain were abdominal/pelvic pain and back/flank pain associated with urinary symptoms, including infections, hematuria, and dyspareunia. Pelvic or abdominal pain presents more frequently than back pain and urinary symptoms (48% vs 30%, p<0.05). 12% of the symptoms were constitutional, with 57% related to weight change, followed by gastrointestinal symptoms (19%), weakness/fatigue/syncope (15%), and night sweats (8%). Weight loss was significantly more common than other individual constitutional symptoms (p<0.01). Vaginal discharge and a deep venous thrombosis were other presenting symptoms. Over half (55%) of the women had two or more presenting symptoms. Of note, ten women had no symptoms at presentation: two of these women had no screening, and both had squamous cell carcinoma stage I. Of the eight women without symptoms or screening, one had adenocarcinoma stage I, and seven had squamous cell carcinoma (six in Stage I, one in Stage III). shows the distribution of type of diagnostic workup by symptom presentation.

Table 3 shows the intersection of screening, symptoms, and diagnostic workup. Of all women (n=126), 34% (43/126) had no screening vs. 66% (83/126) with screening. Likewise, 13% (17/126) of women had no presenting symptoms, and 87% (109/126) had one or more symptoms described in Table 2. Of those with no symptoms (n=17), 76% (13/17) did screen. Among those with no symptoms (n=17), blind biopsy was used significantly more often than colposcopy for the diagnostic workup: 41% (7/17) vs 6% (1/17), p=0.015). Among those with symptoms (n-109), blind biopsy was used significantly more often than colposcopy for the diagnostic workup: 63% (69/109) vs. 10% (11/109), p<0.001). Of those with both screening and symptoms (n=70), blind biopsy was the most frequent diagnostic workup compared to colposcopy: 57% (40/70) vs. 13% (9/70), p<0.001.

Table 3.

Screening, Symptom, and Diagnostic Workup

	No Symptoms	One or More Symptoms	TOTAL
Population Total	17 (13%)	109 (87%)	126 (100%)
No Screening      TOTAL	4 (9%)	39 (91%)	43 (100%)
Blind biopsy	1 (25%)	29 (74%)	30 (70%)
Colposcopy	0	2 (5%)	2 (5%)
Immediate therapy	0	0	0
None	1 (25%)	4 (10%)	5 (12%)
Unknown	2 (50%)	4 (10%)	6 (14%)
Yes Screening      TOTAL	13 (16%)	70 (84%)	83 (100%)
Blind biopsy	6 (46%)	40 (57%)	46 (55%)
Colposcopy	1 (8%)	9 (13%)	10 (12%)
Immediate therapy	4 (31%)	4 (6%)	8 (10%)
None	2 (15%)	13 (19%)	15 (18%)
Unknown	0	4 (6%)	4 (5%)

Greyed rows indicate column-based percentages (e.g., 25% of women with no symptoms and no screening had a blind biopsy for their diagnostic workup.)

^§Blind biopsy was the most frequently used diagnostic workup, regardless of screening or symptoms presentation (60% (76/126) vs. 40% (50/126), p=0.001)).

Table 4 presents the cervical cancer stage by screening and pathology diagnosis. 49% presented in Stage I, 18% in Stage II, 15% in Stage III, and 14% in Stage IV. The majority (71%) of cancers were squamous cell carcinoma, followed by adenocarcinoma (26%). Both cancer types presented in Stage I for about half of the women (49% and 53%, respectively). Adenosquamous and neuroendocrine tumors were rare.

Table 4.

The stage at diagnosis for screening participation and pathology diagnosis

	American Indian	White	Other (Black, Asian, Middle East)	Total
Cervical Cancer Screening	N=25	N=84	N=2	N=111
YES	12 (25%)	36 (75%)	0 (0%)	48 (43%)
Stage I	9 (75%) *	24 (67%) §	0 (0%)	33 (69%) ǂ
Stage II	1 (8%)	3 (8%)	0 (0%)	4 (8%)
Stage III	1 (8%)	6 (17%)	0 (0%)	7 (15%)
Stage IV	1 (8%)	3 (8%)	0 (0%)	4 (8%)
NO	13 (21%)	48 (76%)	2 (3%)	63(57%)
Stage I	3 (23%)	18 (38%)	1 (50%)	22 (35%) **
Stage II	4 (31%)	14 (29%)	0 (0%)	18 (29%)
Stage III	4 (31%)	7 (15%)	0 (0%)	11 (17%)
Stage IV	2 (15%)	9 (19%)	1 (50%)	12 (19%)
Pathology Diagnosis	N=25	N=96	N=3	N=124
Squamous Cell Carcinoma	20 (80%)	65 (70%)	1 (33%)	86 (69%)
Stage I	10 (50%)	31 (48%)	1 (100%)	42 (49%)
Stage II	2 (10%)	13 (20%)	0 (0%)	15 (17%)
Stage III	5 (25%)	8 (12%)	0 (0%)	13 (15%)
Stage IV	3 (15%)	11 (17%)	0 (0%)	14 (16%)
Adenocarcinoma	5 (20%)	27 (28%)	2 (67%)	34 (27%)
Stage I	2 (40%)	14 (38%)	1 (50%)	17 (53%)
Stage II	3 (60%)	4 (17%)	0 (0%)	7 (22%)
Stage III	0 (0%)	5 (21%)	0 (0%)	5 (16%)
Stage IV	0 (0%)	2 (8%)	1 (50%)	3 (9%)
Adenosquamous	0 (0%)	2 (2%)	0 (0%)	2 (2%)
Stage I	0 (0%)	1 (50%)	0 (0%)	1 (50%)
Stage II	0 (0%)	1 (50%)	0 (0%)	1 (50%)
Stage III	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Stage IV	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Neuroendocrine	0 (0%)	2 (2%)	0 (0%)	2 (2%)
Stage I	0 (0%)	1 (50%)	0 (0%)	1 (50%)
Stage II	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Stage III	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Stage IV	0 (0%)	1 (50%)	0 (0%)	1 (50%)

^ǂWomen who participated in screening were significantly more likely to be diagnosed with Stage I than those who were not screened (69% vs. 35%, p<0.001)

^**Women without screening had later-stage cancers significantly more often than women who screened (65% (41/63) vs. 31% (15/48), p<0.001)

^*American Indian women who participated in cervical cancer screening had Stage I diagnosis significantly more often than AI women who did not screen (75% vs 23%, p<0.01).

^§White women who participated in cervical cancer screening had Stage I diagnoses significantly more often than those White women who did not screen (67% vs 38%, p<0.01)

Women who participated in screening were significantly more likely to be diagnosed with Stage I than those who were not screened (69% vs. 35%, p<0.001). Women without screening had later-stage cancers significantly more often than women who screened (65% (41/63) vs. 31% (15/48), p<0.001). Differences were detectable by population: AI women who participated in cervical cancer screening had Stage I diagnosis significantly more often than AI women who did not screen (75% vs 23%, p<0.01). Likewise, White women who participated in cervical cancer screening had Stage I diagnoses significantly more often than those White women who did not screen (67% vs 38%, p<0.01). Supplementary Table 3 presents detailed stage data. Supplementary Table 4 presents the diagnostic workup by symptom presentation.

Supplementary Table 5 shows that nearly all treatment intent was curative (95%) and that the domains of therapy offered were chemotherapy, brachytherapy/external beam radiation, and surgery. Supplementary Table 6 shows the mortality status by screening participation, cancer type, Stage at diagnosis, and number of treatment modalities for each population.

Table 5 shows the predicted modeling of mortality status by age at the time of diagnosis, race of the population, number of presenting symptoms (severity of disease), smoking status (risk factor), insurance status (financial barrier), distance from the treatment center (transportation barrier), screening participation, diagnostic workup modalities, cancer histology, Stage at diagnosis, and number of treatment modalities. The only predictors of survival were younger age at diagnosis, fewer symptoms at diagnosis, and earlier Stage at diagnosis, where only age and Stage I presentation compared to Stage IV remained significant for survival in the adjusted model.

Table 5.

Survival modeling by descriptors

	OR	L95	U95	aOR*	L95	U95
Age at the time of cancer diagnosis	0.97	0.94	1.00	0.97	0.94	1.00
The number of symptoms at presentation *	0.62	0.46	0.85	0.71	0.49	1.02
Distance from the treatment center	1.00	1.00	1.00
Stage at Diagnosis
Stage 1	15.91	4.38	57.72	9.45	2.43	36.71
Stage 2	5.44	1.35	21.88	4.89	1.16	20.63
Stage 3	4.81	1.14	20.25	4.05	0.92	17.85
Stage 4	1.0			1.00
Race
American Indian compared to White	0.75	0.31	1.81
Screening
Screened compared to not screened)	1.88	0.85	4.15
Cancer type
Adenocarcinoma compared to SCC	1.15	0.49	2.70
The number of treatment types
One type	1.00
Two types	0.66	0.29	1.51
Three types	0.63	0.13	3.03
Smoking Status
Never	1.00
Current	0.75	0.31	1.81
Ever	1.03	0.37	2.88
Diagnostic Work-up
Blind biopsy	1.0
Colposcopy+	0.68	0.17	2.78
Excisional therapy	0.26	0.03	2.22
None	1.81	0.67	4.91
Unknown	1.09	0.24	4.91
Insurance Status
Private	2.57	0.64	10.27
Public	1.52	0.40	5.80
None	1.0

The bolded font is statistically significant. Variables were continuous above the line, and categorical below the line.

^*Adjusted for age, number of symptoms, and stage at diagnosis.

Table 6 and Supplementary Figure 2 present the time between the cancer prevention continuum steps. Table 6 presents women stratified by diagnostic workup: colposcopy vs. blind biopsy. Colposcopy found 83% of the cancers in Stage I, whereas blind biopsy found 49% in Stage I, 24% in Stage II, 16% in Stage III, and 11% in Stage IV. Among women whose diagnostic workup started with screening (n=43), no differences were detected between any time-to-event measure. In contrast, among women who presented with at least one symptom (n=109), the time to diagnosis (days) was significantly longer when colposcopy was included in the diagnostic workup (425 (553) vs. 175 (240), p<0.05). However, the mortality rate was not different between the women with colposcopy and those with blind biopsy diagnostic workup.

Table 6.

The time (days) between cancer continuum steps compared by diagnostic workup

	Presenting with screening N=43		Presenting with symptoms N=109
	N	Mean (SD)	N	Mean (SD)
Colposcopy (n=12)
Prior Screen to Diagnosis	9	1457 (1303)	28	5199 (3069)
Screening or Symptoms to Diagnosis	6	440 (653)	8	425 (553) *
Screening to Colposcopy	8	21 (26)	7	15 (21)
Symptoms to Colposcopy	4	453 (697)	6	422 (542)
Colposcopy to First Treatment	9	95 (97)	10	69 (81)
Diagnosis to First Treatment	10	29 (25)	11	23 (23)
First Treatment to Treatment Completion	10	23 (43)	11	21 (41)
Diagnosis to Death	2	1785 (1372)	2	1785 (1372)
Blind Biopsy (n=76)
Prior Screen to Diagnosis	34	3089 (3469)	16	6601 (3885)
Screening or Symptoms to Diagnosis	38	203 (284)	64	175 (240) *
Diagnosis to First Treatment	43	40 (66)	66	32 (54)
First Treatment to Treatment Completion	43	43 (56)	65	53 (54)
Diagnosis to Death	15	788 (740)	25	758 (675)

^*p<0.05

Supplementary Figure 2 shows, on average, 9.2 years (SD 9.5) passed between the last ‘routine’ screening the woman had and her cancer diagnosis. On average, 217 days (SD 277) passed from the time she experienced bleeding, pain, constitutional or other symptoms until her diagnosis. For women with a cervical cancer screen and a colposcopy, the time gap was, on average, 28 days (SD 60). Women with symptoms but no screening took longer to receive colposcopy, on average 162 (SD 271) days (p<0.001). The average time from diagnosis to death among those who died was 3.1 (SD 2.4)) years. Supplementary Table 7 shows the time to death from diagnosis by Stage of disease at presentation.

Discussion

Our data document at the clinical level the details of the continuum of cervical cancer screening, diagnosis, and treatment among AI and White women in the Northern Plains of the US. The only differences we detected between the AI and White women were increased use of public insurance by AI women, residence closer to the treatment center, and a colposcopic step in their diagnostic pathway. AI women with cervical cancer in our study used public health insurance significantly more than AI people in general (42%) as well as the general US population (36%).^1,12

Colposcopic evaluation is not part of the National Comprehensive Cancer Network (NCCN) guidelines¹³ for the workup of a person symptomatic of cervical cancer. In our series, most women were symptomatic, yet 10% had colposcopy. The time from colposcopy to diagnosis in those cases where colposcopy was performed was within the six-month standard of care for asymptomatic women.¹⁴ Bridging guidelines about the need for colposcopy between the American Society for Colposcopy and Cervical Pathology (ASCCP,¹⁵ the American Society of Clinical Oncology (ASCO),¹⁶ and the National Comprehensive Cancer Network (NCCN)¹³ guidelines are necessary. Based on our results, we recommend blind biopsy for women presenting with symptoms of cervical cancer.

Most of the details of our review match with national statistics about women with cervical cancer. Our average age at diagnosis was 53 years compared to 50 years in the general US.¹⁷ In addition, we found a quarter of the cervical cancers in our population were diagnosed at 65 years or older compared to 20% in the general US,¹⁸ highlighting the importance of continuing cervical cancer screening until ten years have passed, during which two consecutive screens indicate no HPV infection or intraepithelial neoplasia ¹⁹. On the other hand, more than a third of our study population used tobacco, significantly more than the general US women’s population of 10%.^20,21 Nonetheless, as is seen in other work, tobacco use did not change the Stage at diagnosis or mortality.²²

Among our study population, only half of the women with cervical cancer had had a cervical cancer screen within the five years before their diagnosis, similar to what is reported nationally.²³ While symptoms are well known, few studies report the detailed frequencies of the presenting symptoms as we do, where intermenstrual/heavy vaginal bleeding was the most common symptom. More than half of the women with cervical cancer have two or more presenting symptoms.²⁴

Our review’s histopathology and Stage at diagnosis differ from national trends,^{13,25, 26,} where our population’s histopathology was nearly 30% adenocarcinoma and almost 70% squamous cell carcinomas compared to a 10%/90% national prevalence.²⁷ 47% of our adenocarcinoma diagnoses were made in late stages, and 70% of our population in late-stage adenocarcinoma died, both higher frequencies than others found.²⁸ It is implausible that biological differences in our population would account for these differences. However, less than half of the women with adenocarcinoma participated in screening.

In our population, women beyond the screening age (65 years) presented more often in the early stages, whereas national trends indicate older women present in later stages.²⁹ While the distribution of Stage at diagnosis among our population differs from national trends, it does agree with the regional trends identified by the Centers for Disease Control and Prevention (CDC).³⁰ Most agree that screening cessation is not just an age achieved but also a ten-year history of no abnormalities, a history that unscreened women do not have.

The treatment types and number of combinations of treatment types are consistent with national recommendations.¹³ Mortality in our population was also similar to national trends at 64% ¹⁷ but was not different by population subgroups,³¹ in part because our study was not powered to see differences. Among other researchers,^32–34 cervical cancer mortality from 1999–2009 was twice as high among AI as NHW (4.2/100,000 AI women against the US national rate of 2.1/100,000), where AI women 65–84 years had a 10/100,000 mortality rate (2.8 fold higher than NHW). Those older than 84 had a 23.7/100,000 mortality rate (6.1 fold higher than NHW).⁶ Others have shown that AI of the Northern Plains had the highest relative mortality (4.2 compared to NHW) among all tribal communities.⁶ Through 2019, mortality from cervical cancer remained 64% higher among AI than among NHW women and 2.9 fold higher for AI women 50–64 years than the NHW 50–64-year-old women.^7,8

A unique offering of our work is the detailed time-to-event data for this population,^17,35,36, much of which is not available in the literature. We showed the need for an improved expedited diagnostic workup. Current 2019 ASCCP guidelines encourage. women whose screening result is HPV 16 positive and a high-grade squamous intraepithelial lesion (HSIL) cytology to have expedited tissue diagnosis. But, in our population, where screening was not common, it was instead the need for expedited tissue diagnosis among women with irregular heavy bleeding and/or abdominal or back pain.

Strengths and Limitations

The strength of this work is the documentation of the cervical cancer prevention continuum among a large portion of AI women living in the same catchment as White women, all of whose cervical cancer was treated at the same cancer center. We provide the highest level of detail of clinical information describing the cervical cancer prevention continuum among all publications to date. In addition, while we are not powered to see differences between the two population groups, having a group of White women from the same catchment area as a comparator may indicate that the failures in the cancer prevention continuum are less race-associated than overall discontinuities in health care information sharing worsened by the lack of integration of all health care systems.

The limitations of this work include the source of data, which, while accurate for medical care, could not include all data elements for all patients, as is common with registry-based data analyses.³⁷ Additionally, this is a review of cervical cancer patients and hence does not represent the population view of screening or diagnostic workups.

Conclusions.

Women symptomatic for cervical cancer at presentation must be evaluated for immediate histologic diagnosis.

Abbreviations and Acronyms

AI

American Indian

ASCCP

American Society for Colposcopy and Cervical Pathology

ASCO

American Society of Clinical Oncology

CDC

Centers for Disease Control and Prevention

FIGO

International federation of obstetrics and gynecology

HPV

human papillomavirus

HSIL

high grade squaous intraepithelial lesion

IRB

Institutional Review Board

NCCN

National Comprehensive Cancer Network

NHW

non-Hispanic White

PRCDA

purchased/referred care delivery care

UIHO

Urban Indian Health Organization

US

United States