Article Text
PDF
Abstract
Objective To estimate the predictive role of faecal haemoglobin (f-Hb) concentration among subjects with faecal immunochemical test (FIT) results below the positivity cut-off for the subsequent risk of advanced neoplasia (AN: colorectal cancer—CRC—or advanced adenoma).
Design Prospective cohort of subjects aged 50–69 years, undergoing their first FIT between 1 January 2004 and 31 December 2010 in four population-based programmes in Italy.
Methods All programmes adopted the same analytical procedure (OC Sensor, Eiken Japan), performed every 2 years, on a single sample, with the same positivity cut-off (20 µg Hb/g faeces). We assessed the AN risk at subsequent exams, the cumulative AN detection rate (DR) over the 4-year period following the second FIT and the interval CRC (IC) risk following two negative FITs by cumulative amount of f-Hb concentration over two consecutive negative FITs, using multivariable logistic regression models and the Kaplan-Meier method.
Results The cumulative probability of a positive FIT result over the subsequent two rounds ranged between 7.8% (95% CI 7.5 to 8.2) for subjects with undetectable f-Hb at the initial two tests (50% of the screenees) and 48.4% (95% CI 44.0 to 53.0) among those (0.7% of the screenees) with a cumulative f-Hb concentration ≥20 µg/g faeces. The corresponding figures for cumulative DR were: 1.4% (95% CI 1.3 to 1.6) and 25.5% (95% CI 21.4 to 30.2) for AN; 0.17% (95% CI 0.12 to 0.23) and 4.5% (95% CI 2.8 to 7.1) for CRC. IC risk was also associated with cumulative f-Hb levels.
Conclusion The association of cumulative f-Hb concentration with subsequent AN and IC risk may allow to design tailored strategies to optimise the utilisation of endoscopy resources: subjects with cumulative f-Hb concentration ≥20 µg/g faeces over two negative tests could be referred immediately for total colonoscopy (TC), while screening interval might be extended for those with undetectable f-Hb.
- FIT
- CRC screening
- advanced adenoma
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
A correlation of faecal haemoglobin (f-Hb) concentration at first screening tests with the subsequent risk of colorectal cancer (CRC) and advanced adenomas (AAs), has been documented. Two studies also shown an association of cumulative f-Hb concentrations in predicting the subsequent risk of CRC among screenees with two negative faecal immunochemical test (FIT) examinations.
What are the new findings?
The positivity rate and advanced neoplasia (AN—CRC + AA) detection rate at subsequent screening tests, as well as the interval CRC rate, are associated with the f-Hb level at previous tests. Considering the first two FIT exams, subjects with cumulative f-Hb concentration ≥20 µg Hb/g faeces are showing a very high risk of being detected with AN at subsequent rounds, while the subsequent AN risk is very low among subjects with undetectable f-Hb levels. A similar trend could be observed for the IC risk.
Significance of this study
How might it impact on clinical practice in the foreseeable future?
The findings of our study provide information, which could be relevant for defining tailored protocols following the initial screening rounds. A longer screening interval might be considered for low-risk (LR) subjects with undetectable f-Hb levels at the first two FIT exams, while immediate TC referral at the second FIT of subjects with the highest cumulative f-Hb concentration might be justified. Such approach might result in a more appropriate and more efficient utilisation of endoscopy resources, while reducing unnecessary assessments among LR subjects.
Introduction
Faecal immunochemical tests (FITs) have been recommended as the faecal test of choice for population-based screening programmes.1
One of the main advantages of FITs is that they allow for faecal haemoglobin (f-Hb) quantification. Several reports2–5 have documented a positive correlation between f-Hb concentration and those morphological and histological characteristics associated with the severity of colorectal neoplasms detected at colonoscopy (TC), among subjects with a FIT result exceeding the positivity threshold. While an overlap between individuals with a negative TC and those with a low-risk (LR) adenoma was observed, high-grade dysplasia, villous histology, distal location, increasing size and number of adenomas were all features associated with an increased f-Hb level.2–5
In the context of population-based programmes, quantitative information is usually used to modulate the positivity cut-off, to match the local endoscopy capacity, but the test results are then used in a dichotomised fashion. The quantitative nature of FIT offers, however, new options for screening, which have not been fully explored. Evidence from studies conducted among subjects with positive FIT results is suggesting that combining f-Hb level and individuals characteristics may allow to define subgroups of subjects showing a different risk of being detected with advanced adenomas (AAs), or colorectal cancer (CRC).6 The positive predictive value (PPV) for advanced neoplasia (AN: AA or CRC) was ranging between 24% among women aged 50–59 years with 20–64 µg Hb/g faeces and 76% among men aged 60–69 years and f-Hb level >175 µg/g faeces in an average risk screening population in Spain.6
The predictive role of f-Hb concentration among subjects with FIT results below the positivity cut-off has been assessed in three recent studies, conducted in the context of mass screening programmes. The f-Hb concentration, at first screening,7 or over two consecutive negative FITs8 9, emerged as an independent predictor of the subsequent risk of AN, both screen detected (SD) and not SD, although only one study9 could provide information about the association of f-Hb concentrations of consecutive tests with the screening outcomes at the subsequent round.
The aim of this study was to estimate, among subjects with previous consecutive negative FITs, the risk of being detected with an AN, or of being diagnosed with an interval CR (IC), following a positive result at subsequent tests, by f-Hb concentration at previous negative examinations.
Methods
The study was conducted in the context of four population-based screening programmes in Northern-Central Italy: the Piedmont Region programme, targeting all residents aged 59–69 years, the Aosta Valley programme, targeting people aged 50–74 years, and the Florence and Reggio Emilia programmes, targeting people aged 50–69 years. All eligible subjects are invited every 2 years to perform a single sample FIT, with a personal invitation letter, signed by their GP. A reminder letter is sent to all non responders to the initial invitation. Participation rates in each screening round ranged between 44.7% and 67.7% during the reference calendar period of the study.10
We included in the analysis subjects aged 50–69 years, who performed their first screening FIT between 1 January 2004 and 31 December 2010 and had one, two or three additional consecutive tests following an invitation issued within December 2014. Subjects performing two consecutive tests with an interval shorter than 18, or longer than 60 months, were excluded, as they might represent self-selected subgroups of subjects, responding after two previous consecutive refusals, or requesting to anticipate the next examination, following onset of symptoms. Eligible subjects with two consecutive negative tests were followed-up over the subsequent two rounds to estimate their likelihood of having a positive result at the third and fourth round and of being detected with AA, or CRC, by cumulative f-Hb concentration at the initial two negative FITs.
Subjects with negative FIT result were informed by mail that their test was negative (no mention of the f-Hb concentration) and that they would have been invited again after 2 years; those with positive FIT results were contacted by the screening staff and they were offered an appointment for a TC. The examination was performed in the endoscopy reference centres for each programme, using standard colonoscopes, following a cathartic bowel preparation. TC was considered complete if the caecum was visualised or, in the case of failure, when a subsequent exam, performed within 6 months, reached the caecum; the combined results of the two examinations have been considered for analysis in these latter cases.
All SD lesions were annotated according to size, morphology and localisation. Histology was defined according to the WHO criteria11: AA was defined as an adenoma with any of the following features: size ≥10 mm, high-grade dysplasia or villous component >20%; cancer was defined as the invasion of malignant cells beyond the muscularis mucosae.
Data about ICs (including CRCs in screenees who were no longer eligible for, or who did not respond to, the subsequent invitation) were available from three centres (Turin, Aosta Valley and Reggio Emilia) through a record linkage of the study database with the local population cancer registries, which were up-to-date until 31 December 2012.
During the reference period of this study, all programmes were using the same analytical procedure (OC Sensor, Eiken Japan) and they had adopted the same positivity cut-off, set at 20 µg Hb/g faeces (100 ng/mL buffer). The screening test results were dichotomised as positive or negative, but the results of the f-Hb determinations were recorded in the screening database in all centres. Although the reliability of the measure could not be considered accurate enough to provide a valid estimate for very low levels of f-Hb, numerical data in the range 1–9 µg Hb/g faeces were available over the period covered by this study from all laboratories, but Reggio Emilia, where subjects with values <10 µg Hb/g faeces were classified in the group with undetectable f-Hb, starting from March 2010. All laboratories are regularly involved in an external quality assurance (EQA) scheme operating according to ISO 17043, supported by the reference laboratory from Tuscany Region.12
Screening activity within these programmes is monitored on regular basis, using a standard common set of quality indicators.13 We conducted this analysis as a part of the routine monitoring activity of the programmes performance, pooling anonymous individual data from each programme, based on a common standardised form. Approval from local ethics review boards is not required for monitoring programme activity.
Statistical analysis
We conducted a preliminary analysis aimed to assess the potential correlation of the results of repeated tests from the same subject. Although statistically significant, the correlation between f-Hb concentrations at different tests was low (first vs second: ρ=0.13; second vs third: ρ=0.10; first vs third: ρ=0.08). Also, the f-Hb concentration at each test emerged as an independent predictor of the outcomes of interest (positivity rate (PR) and AN detection rate (DR)), in a multivariable logistic regression model, adjusted by age, gender, interval between the tests and screening programme; the association was stronger for the last test result, while the interaction terms for f-Hb levels were not significant. The same trend could be observed both when using absolute f-Hb values and their log transformation.
Based on these results, suggesting a weak correlation of f-Hb concentrations, we categorised screenees using two approaches: (a) the total amount of f-Hb detected over the previous tests (ie, the sum of the f-Hb concentration at previous tests), classifying subjects into six groups : (1): 0; (2): 0.1–3.9; (3): 4–9.9; (4): 10–14.9; (5): 15–19.9; (6): ≥20 µg Hb/g. faeces; (b) the weighted average of the log of f-Hb concentration at the previous test, again classifying subjects into six groups, based on the percentiles of the distribution (0; median, 75, 90, 95, 99 percentiles).
To assess the association of the variation in the amount of blood loss over time and the AN risk at the third/fourth round, we also classified subjects based on the observed change in the f-Hb concentration from the first to the second FIT, using separate classes for negative and positive change: no change; change from one group to the nearest one; change ≥2 groups (ie, from group 1 to group 4).
Subjects with a positive test result at the third FIT were excluded from the analysis of the results of the subsequent round, independent of the result of TC referral. We compared PR, DR of AA and CRC, AN PPV and number needed to scope (NNScope) in the f-Hb concentration categories and we ran multivariable logistic models to estimate the association between the level of cumulative f-Hb concentration over two tests and the AN DR at subsequent rounds, adjusted by the interval since the last negative FIT (months: 18–22; 23–27; 28–32; 33–36; 37–60), gender, age (5-year age groups) and screening programme, to allow for variability in the background incidence rates. We then derived, based on the adjusted ORs, the predicted probability of being detected with an AN at the third round by gender, age and cumulative f-Hb level, for subjects attending at the standard screening interval (22–27 months).
Cumulative FIT positivity, as well as CRC and AN DR, were estimated using the Kaplan-Meier method for subjects with two negative FITs who attended in the third and fourth round. Each subject contributed to the person-time with the interval between the dates of their second negative FIT and of their last FIT (third or fourth exam), or last TC (if last FIT was positive).
We computed the cumulative IC risk and the average rates per 100 000 person-years deriving their 95% CIs by using the standard error (SE) of the log transformation of the rates.14
The exact method was used to calculate the 95% CIs of proportions; 95% CI for the NNScope were calculated as the inverse of the 95% CI of the corresponding PPV.
All statistical tests were two-sided and statistical significance was set at 0.05.
Results
Out of 290 038 subjects having their first negative test between 1 January 2004 and 31 December 2010, who were eligible (subsequent invitation date within 31 December 2014 and age at subsequent invitation <70 years) for a second invitation 256 949 (88.6%) underwent a second FIT (figure 1): in this subgroup 89.9% (165 681/184 373) of those eligible for 3 invitations had 3 tests and 91.7% (80 957/88 332) of those eligible for 4 invitations had 4 tests.
Screenees’ flow. AA, advanced adenoma; AN, advanced neoplasia; CRC, colorectal cancer; DR, detection rate; FIT, faecal immunochemical test; PPV, positive predictive value.
The cumulative f-Hb over the initial two tests, among subjects undergoing their third FIT, was <10 µg/g faeces in 94.2% of the screenees and over the predefined positivity cut-off of 20 µg/g faeces in 0.7% (table 1). The PR and the AN PPV among subjects performing their third FIT showed a strong positive correlation with the cumulative amount of f-Hb after two tests: the PR ranged between 2.6% and 25.9% among subjects in the lowest as opposed to those in the highest cumulative f-Hb category; the AN PPV, ranging between 16.3% and 50.6%, as well as the DR for CRC, ranging from 0.04% to 1.86%. and for AA, ranging from 0.3% to 9.8%, showed a parallel increasing trend.
Association of cumulative f-Hb values over the initial two rounds and FIT PR, AN PPV, AA and CRC DR, NNScope to detect 1 AN at the third round - Men and women aged 50–69 years
These same correlation could be observed among men as among women and in the younger, as in the older age groups. The overall PR, PPV and AN DR were higher among men than among women, showing a trend towards an increase with age (online supplementary appendix tables 1A-1B).
Supplemental material
Both CRC and AA DR at the third screening test (table 2) were higher among men than among women and they were lower in the youngest age group, as compared with subjects older than 64; an increase (OR: 1.49; 95% CI 1.14 to 2.00) in the likelihood of being detected with an AA was observed among subjects who had performed their last negative FIT since >3 years, while there was no association of interval since last FIT and CRC risk. Adjusting for these variables, the CRC and AA DR maintained a strong correlation with the cumulative f-Hb concentration over the previous two exams: as compared with subjects with non detectable f-Hb at previous exams, the likelihood of being detected with a CRC was twice as high (OR: 2.26; 95% CI 1.47 to 3.46) among subjects with f-Hb <4 µg/g faeces and it showed a 12-fold (OR: 11.63; 95% CI 6.42 to 21.07) and 39-fold (OR: 38.92; 95% CI 22.50 to 67.31) increase among subjects with f-Hb levels between 15 and 19.9 µg Hb/g faeces and ≥20 µg Hb/g faeces, respectively. The corresponding ORs for AA were 1.75 (95% CI 1.47 to 2.07), 12.84 (95% CI 10.32 to 16.00) and 30.40 (95% CI 24.09 to 38.38).
Predictors of the DR of CRC, advanced adenoma and AN at the third FIT
Among subjects showing a substantial decrease of f-Hb concentration at the second as compared with first FIT (from f-Hb >15 to f-Hb <4 µg/g faeces) the likelihood of being detected with AN at the third test, although reduced, remained ninefold higher, as compared with subjects showing f-Hb levels <4 µg/g faeces at both exams (data not shown).
The same trends could be observed when classifying subjects undergoing their third test based on the weighted average of the log of the two previous f-Hb determinations, as well as when restricting the analysis to subject with f-Hb concentration <15 µg/g faeces at the first FIT, simulating the adoption of a lower positivity cut-off (data not shown).
The proportion of SD proximal CRCs (table 3) tended to be higher among screenees with a positive result at the third FIT than among those testing positive at the second test (45.2% vs 37.8%; OR: 1.20%; 95% CI 0.97 to 1.47) and, among screenees testing positive at the third FIT, it was higher among those with detectable cumulative f-Hb <10 µg/g faeces than among those with cumulative f-Hb ≥20 µg/g faeces (57.1% vs 28.6%; OR: 3.33%; 95% CI 1.09 to 11.36). No such trend could be observed for AA.
Site distribution of SD CRCs and advanced adenomas by cumulative f-Hb values over the initial two rounds and among subjects with a positive result at the second FIT
Union for International Cancer Control (UICC) stage classification was available for 75% of SD CRCs: the proportion of advanced (UICC III or IV) CRCs (29.3% of 140 classified CRCs) did not show an association with the cumulative f-Hb concentration, or with the interval since last negative FIT.
The predicted AN DR at the third round (figure 2), among subjects attending screening at the standard 2-year interval, was <0.5% among women with undetectable f-Hb at previous exams and 15.8% among older men with cumulative f-Hb ≥20 µg/g faeces.
Predicted an DR by age and gender by cumulative f-Hb at the second FIT. DR, detection rate; f-Hb, faecal haemoglobin; FIT, faecal immunochemical test.
The cumulative f-Hb concentration over the initial two FITs showed the same positive association with the IC risk over the 36 months following the second negative test as with the risk of being detected with a CRC at the third round (table 4). The cumulative incidence of CRC (IC + SD CRCs) over the 3-year period following the second FIT was significantly lower among subjects with undetectable f-Hb at previous test as compared with those with cumulative f-Hb >0 µg Hb /g faeces (OR:0.26; 95% CI 0.17 to 0.41). The same trend could be observed when restricting the analysis to a 24-month follow-period (data not shown).
CRC DR at the third FIT* and IC risk during the 36 months following second negative FIT by cumulative f-Hb concentration over two consecutive FITs—screenees from three centres
The predictive role of the cumulative f-Hb concentration was maintained also among subjects undergoing their fourth exam, when classifying subjects based on the cumulative f-Hb concentration over three tests (online supplementary appendix table 2). The cumulative f-Hb concentration over the initial two tests was still predictive of the screening outcomes at the fourth round, although the association was stronger when using the cumulative f-Hb concentration over three tests, after adjusting for age, gender, interval since the third FIT and screening programme (online supplementary appendix table 3).
The cumulative probability of having a positive FIT result over the subsequent two rounds (within 54 months since the second negative FIT) ranged between 7.8% (95% CI 7.5 to 8.2) for subjects who had no f-Hb detected at the initial two tests and 48.4% (95% CI 44.0 to 53.0) among those who had a cumulative f-Hb concentration exceeding 20 µg Hb/g faeces (data not shown).
The corresponding figures for cumulative DR were 1.41% (95% CI1.3 to 1.6) and 25.5% (95% CI 21.4 to 30.2) for AN (table 5); 0.17% (95% CI0.12 to 0.23) and 4.5% (95% CI 2.8 to 7.1) for CRC (data not shown).
Cumulative DR of an (advanced adenoma or CRC) by cumulative f-Hb concentration over two consecutive FITs and time since the second FIT
The cumulative AN DR was similar among subjects younger than 60 as among those aged 60–69 years, while it was higher among men, as compared with women, in all age groups (online supplementary appendix table 4).
The cumulative AN DR was the same in the group with highest cumulative f-Hb (24.3%; 95% CI 19.5 to 30.0), when excluding from the analysis the exams performed in Reggio Emilia after March 2010 (online supplementary appendix table 5), while it was lower (1.0%; 95% CI 0.9 to 1.2) among subjects with undetectable f-Hb at the previous two FITs, as compared with the results of the main analysis.
Assuming that SD AN at the third, or fourth, round could be detected already at the time of the second FIT, anticipating TC referral of subjects with cumulative f-Hb levels ≥20 µg Hb/g faeces at the second FIT would result in an AN yield at TC achieving 25.5% (95% CI 21.4 to 30.2), slightly lower than the observed 30.1% (95% CI 29.1 to 31.1) AN PPV of a positive FIT at the second examination (figure 1). Taking into account the estimated 48% cumulative PR in this subgroup, this approach would result in a 9% increase in the TC workload over the subsequent two rounds. Extending the screening interval in the group showing the lowest AN risk, the expected reduction in the TC workload would achieve 12%, when considering a 6-year period (ie, comparing two vs three screening rounds).
Discussion
This study, conducted in the context of large population-based programmes, showed a strong association of the cumulative f-Hb concentration over the first two negative screening tests with the probability of having a positive result and of being detected with an AN at subsequent examinations. A similar trend could be observed for SD CRCs as for IC risk in the 36-month period following the second negative FIT.
In particular, we identified two subgroups at different risk: as compared with those with undetectable f-Hb at previous FITs (about 50% of the screenees), showing a 1.4% risk of being detected with an AN over the subsequent two rounds, subjects with a cumulative f-Hb level ≥20 µg/g (about 0.7% of the screenees) showed a 18-fold increase in their cumulative AN risk (26 times for CRC) over the same interval. Their risk was still twice as high as compared with the subjects with a cumulative f-Hb in the range 15–19.9 µg/g. The cumulative AN risk was similar in the younger (50–59) as in the older (60–69) age group, while it was higher among men, as compared with women, as already reported in a previous analyses.8 9
Within-subject correlation between consecutive f-Hb levels was weak, although statistically significant, and, as already reported,8 9 the predictive value of cumulative f-Hb was similar, independent of the sequence of the f-Hb results.
The parallel increase in the PPV for CRC and AA, following a positive result at subsequent exams, would suggest a positive association of f-Hb levels with the subsequent risk of AN. This association could be observed either when measuring the cumulative f-Hb as the sum of the concentrations detected at previous exams, or when considering the weighted average of the log transformation of previous f-Hb values. The sum of f-Hb concentrations over previous rounds would thus seem the preferred parameter to be adopted, when using f-Hb values from previous negative test to tailor subsequent screening intervals and TC referrals, as it can be easily derived from the information routinely available from population-based programmes databases.
Of note, even if a previous health technology assessment report15 suggested that the 4 µg Hb/g faeces value may represent the detectability threshold of the method, also subjects totalling cumulative a f-Hb concentration in the range 1–4 µg/g faeces showed a twofold and 75% increase in the risk of being detected at the third round with a CRC and an AN, respectively, as compared with screenees with undetectable f-Hb.
Previous studies had explored the predictive role of f-Hb using the information from a single exam. Lowering the positivity threshold for the FIT has been shown to be associated with an increase in sensitivity, in particular for AA16 in population-based screening, although at the cost of reduced specificity,17 and in a decrease in the AN DR at the subsequent round.16
Other studies7 8 had documented a strong positive association of baseline f-Hb concentrations below the stipulated positivity cut-off with the long-term risk of AN, with a reported eightfold increase at 8-year follow-up (four screening rounds) among screenees with f-Hb ranging between 8 and 10 µg Hb/g faeces at the first FIT, as compared with those with undetectable f-Hb in the Dutch study.8
Evidence supporting the predictive role of repeated f-Hb determinations over consecutive rounds, among screenees with FIT results below the stipulated positivity cut-off, is instead limited.
The Dutch study8 reported, over a 8-year follow-up, a similar trend in the overall AN risk, including both SD and ICs, by cumulative f-Hb after two consecutive negative FITs as in our study, but it did not provide specific information by screening round, or about the size of the risk subgroups, which might be useful to orient screening policies. A recent Spanish study,9 using the same cut-off as in our programmes, reported a similar 22-fold increase in the risk of being detected with an AN at the third round among subjects with two consecutive high (10–19.9 µg Hb/g faeces) negative FITs. However, the choice to classify subjects based on nine possible combinations of three broad groups of f-Hb concentration may not adequately account for the available quantitative information and it may not be easily translated into practice, as there are three combinations showing a similar risk. Also, the inclusion of subjects with very low (1–3.8 µg/g) f-Hb concentration in the non-detectable group would result, based on our findings, in a misclassification of subjects at LR.
The findings of our study, extending the follow-up to the fourth round, provide additional information, which could be relevant for defining tailored protocols.
The high AN risk atsubsequent tests, together with the higher risk of IC within the 36-month period following the second test, in the small group of subjects with cumulative f-Hb ≥ 20 µg Hb/gr. faeces over the initial two rounds would justify a policy stipulating immediate TC referral. Assuming that all ANs detected over the subsequent 2 rounds would have been detected when performing a TC at the time of the second exam, anticipating the TC referral in this sub-group might represent a cost-effective approach, still ensuring a satisfactory PPV of TC referral, while possibly achieving a more favourable stage distribution of SD CRCs, as well as of ICs, now detected with 2-4 years delay, with a limited increase in the endoscopy workload. Also, as the observed trends were maintained also when restricting the analyses to subjects with f-Hb concentration at the first round below 15 µg/gr. faeces (i.e. assuming to adopt a lower cut-off at the first round), such strategy might be considered independent of the positivity cut-off adopted. Based on our findings, such approach would not improve, however, the proximal AN yield ofthe third FIT, as proximal SD CRCs showed a low-cumulative f-Hb concentration over the previous two rounds, while the proportion of SD proximal AA was not associated with the cumulative f-Hb concentration.
Subjects with undetectable f-Hb at previous exams, making up about 50% of the screenees, showed a very low cumulative AN (1.41%) and CRC (0.17%) DR at 54 months since their second FIT and they might therefore be offered screening at longer interval. The very low CRC (SD and IC) risk among these subjects would justify the implementation of a comparative study aimed to assess the burden of false positive results, as well as the IC rate, when using extended screening intervals. The adoption of a longer screening interval might show alimited impact, as these subjects might be more likely to harbour non-bleeding lesions. Available evidence from a modelling study18 is indeed supporting the hypothesis that a certain proportion of adenomas and CRCs are missed by repeated FIT screening,due to systematic false-negative results, presumably caused by non bleeding neoplasms.
We included in the main analysis also screenees from the centre where f-Hb levels in the range 1–9 were classified as undetectable f-Hb, starting from 2010. However, as we restricted the analysis to subjects performing their first FIT within 2010, the cumulative amount of f-Hb over two determinations could be underestimated only for about 10% of the cases. The observed trends among subjects with detectable f-Hb were indeed maintained when excluding this centre from the analysis, while the cumulative AN DR was significantly lower among subjects with undetectable f-Hb, which would suggest that avoiding misclassification of subjects with very low f-Hb, might improve the discriminative ability of the cumulative f-Hb determination, achieving a more accurate classification of LR individuals.
Data about IC incidence were available only for a subgroup of screenees and the information was updated only up to 2012, which did not allow to quantify the CRC risk over several rounds. Our results are however confirming the findings of previous studies9 showing a similar positive association of cumulative f-Hb with the risk of SD AN and of IC. Also, the IC rate observed in this subgroup is consistent with the estimates form previous reports in the areas covered by the study19 20
The lack of information about lifestyle-related risk factors, anthropometric measures and family history, possibly contributing to a more accurate risk stratification of screenees21, represents an additional limitation of our analysis, although it should be considered that such information would not be routinely available in a population-based screening setting and that the cost-effectiveness and feasibility of approaches combining FIT results and lifestyle-related risk score need to be evaluated.
We did not observe any change in the observed trends, when simulating the adoption of lower positivity cut-off at the initial round. Also, the results of the analysis of the association between f-Hb concentrations below the positivity cut-off (80 µg/g) used in the Scottish22 screening programme and the outcomes in the subsequent round are consistent with our findings.
However, the generalisability of our conclusions to settings using different kits and/or adopting different positivity thresholds, or screening intervals, may require further assessment. In particular, changes in the specimen collection, as well as in the composition of the buffer, introduced in the most recent years to make f-Hb less susceptible to the influence of external physical factors, documented in previous studies23 24 might influence the distribution of f-Hb concentration at each single determination and consequently its cumulative distribution over several tests. A recent study, within a population-based programme, showed indeed that f-Hb was more stable when faecal samples were collected with devices using the new buffer formulation, particularly in those samples with a smaller quantity of f-Hb, as compared with those with higher concentrations, resulting in a reduction of the proportion of subjects with undetectable f-Hb.25 Although the PR was similar with the new as with the old collection device, the proportion of subjects showing a cumulative f-Hb above the positivity threshold would likely tend to increase, with a parallel decrease in the proportion of subjects with undetectable f-Hb at repeated tests. While this trend might improve the negative predictive value of repeated negative FIT determinations, it might result in a reduction of the PPV among subjects in the highest f-Hb concentration group.
In conclusion, the AN prevalence at subsequent screening is associated with the f-Hb level at previous tests: subjects with cumulative f-Hb concentration ≥20 µg Hb/g faeces are showing a very high risk of being detected with AN at subsequent rounds, while the subsequent AN risk is very low among subjects with undetectable f-Hb levels at the first two FIT exams. A longer screening interval might be considered in this latter group, while immediate TC referral at the second FIT of subjects with the highest cumulative f-Hb concentration might be justified, as it might result in downstaging, or preventing, those CRCs now detected 2, or 4, years later. The increase in the TC demand at the second round might be balanced by its decrease in the subsequent rounds, as a result of the adoption of longer screening intervals for subjects with undetectable f-Hb. Such an approach might result in a more appropriate and more efficient utilisation of endoscopy resources, while reducing unnecessary assessments among LR subjects. Sequential randomised trials would be needed to confirm these hypotheses. Effective communication strategies should be developed as well to adequately convey the information about risk tailored screening intervals among subjects with negative FIT results.
Acknowledgments
We would like to acknowledge the contribution of the Local screening programmes coordinators (Dr MP Alibrandi, Dr M Sartori, Dr F Germinetti, Dr P Bestagini, Professor C Magnani, Dr T Miroglio, Dr G Faragli) as well as of the laboratory technicians, the endoscopists, the pathologists, as well as of the screening staff involved in the delivery of screening in the participating centres. We would like to thank Dr Guglielmo Ronco (CPO—Piemonte) for statistical advice.
References
Footnotes
Correction notice This article has been corrected since it published Online First. The author affiliations have been amended.
Contributors Study concept and design: CS, MZ, NS and GG. Acquisition of data: AA, CV, FG, LO, MF, PC, PM, RC, RL, SC, SP, RS. Analysis and interpretation of data: CS, MZ, NS, CC, GG, TR and SR. Drafting of the manuscript: CS, TR, SR, CV and CC. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: CS, PA and ER. Study supervision: NS, MZ and CS.
Funding The study was conducted in the context of the routine monitoring activity of programmes performance.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.