top of page
Search

Decloaking the Malicious Walnut

I've always relished food, especially when a food item becomes associated with a common body structure. In this case, the prostate. The first time I've ever heard about the prostate was in a year 1 anatomy class, where I studied the effects of hormonal derangements in men as well as the role of the prostate in sexual function. The lecturer described it as 'walnut-shaped'. I guess it just got stuck in my head since then.


Scant studies have been performed regarding the detection of a bad 'nut', i.e. when the walnut gets malicious (here, I'm referring to the prostate, not the testicle which is often known as a 'nut' in colloquial circles; I'll talk about testicular cancer on another day). Traditional diagnostic methods in place for prostate cancer are digital rectal examination (where the examiner inserts a finger in the rectum to assess the consistency and size of the prostate by palpating the anterior aspect of the back passage; it's actually worse for the examiner than the examinee...) and ultrasound, where the contours of the structure are mapped. If anything weird is detected in the process, a biopsy is performed. Prostate tissue is collected for pathological assessment, aka looking under the microscope to see if there are any abnormal cells around. However, current methods are not perfect. Prostate cancer is slow-growing and it is more common for the patient to die with the disease, rather than die of it. Autopsy studies have shown high rates of undiagnosed prostate cancer in men, where it has been estimated that 42 to 66 per cent of prostate cancers cause no clinical harm to the patient if they remain undetected. [3] Current methods lead to the over-detection of cancerous lesions which are not aggressive. In other words, they are 'clinically insignificant'. Apart from leading to heightened 'anxiety' and hyper-surveillance, this entails unnecessary and overly radical treatment. In the long run, there is thus higher risk for the patient to die of the side effects of treatment rather than the cancerous lesion itself. [4] It is reasonably understandable why doctors administer treatment - patients are anxious about their health and the social connotations of cancer dictate that no cancer should remain in the body, even the indolent ones. It is very difficult to insist not to do anything when the patient is alarmed by being informed of having a malicious walnut in his body.

Illustration of Digital Rectal Examination (extracted from: https://medipresskenya.com/wp-content/uploads/2019/12/DRE.jpg)


The level of aggressiveness of a lesion is determined by the Gleason Score, which is calculated by the addition of two parts: (a) primary, or predominant grade, and (b) secondary, or highest non-predominant grade. The maximum score for any individual component is 5, thus giving a total, maximum score of 10. Although the technical range is 2-10, it is conventional practice for pathologists to give a score of 6-10 (since there must be a striking abnormality to warrant a biopsy being done; no asymptomatic man would volunteer to do a biopsy). For a lesion to be clinically insignificant, the Gleason Score must be 6. Clinically significant lesions, following this, have Gleason scores within the range 7-10. An example would be 4+3, giving a total score of 7. [5]


What we're looking for in a diagnostic modality, at the end of the day, is heightened detection rates of clinically significant cancers, decreased detection rates of clinically insignificant cancers, and fewer instances necessitating biopsy (which is triggered by an abnormal result on the scan). It follows that the more sensitive and specific the modality is, it is more likely to be reflective of the low-grade status of the cancer and a biopsy is not unnecessarily performed. Furthermore, following advancements in radiology, different diagnostic modalities come with different side effects. What we want is reduced side effects. Giving general examples, X-Ray and CT require the use of radiation. MRI and ultrasound, on the other hand, do not. Contrasts may also be required, depending on the circumstances. For instance, when identifying tumours in the brain, administering contrast can improve diagnostic accuracy - both in determining whether it's a tumour in the first place, and what type of tumour that is. MRI and CT contrasts are understandably different. MRI uses a gadolinium-based contrast, which has fewer side effects than the iodine-based counterpart in CT, especially those occurring in the kidneys. [6]

7.0T MRI Brain (T1 sequence; Coronal [left] and Sagittal [right] views) (extracted from: https://aehrc.com/wp-content/uploads/2016/06/MP2RAGE.png)


There has been much generated interest in the use of MRI in prostate cancer detection. Watching the lecture on prostate cancer diagnostics and future research directions hosted by UCL [1], one cannot help wondering whether MRI is the game-changer we all needed. MRI can be used as not only a diagnostic tool, but a tool for administering treatment. Before we get ahead of ourselves, we should return to the core question: should we use MRI instead of the current protocol (ultrasound) in diagnosing prostate cancer? For long, this question is explored only by single-centre studies. As we all know, bigger is better and this applies to studies as well. We want as diverse and large a cohort as possible. If the same results can be replicated in a much larger study, it means the results are more reliable and trustworthy. Now, the PRECISION Study from UCL has answered the question. Here, I explore in the following their paper published on the New England Journal of Medicine. [2]


The Paper


Before I begin, I have to clarify that this study is slightly different from single-centre studies reported previously. Those compare (a) with (b), where (a) = MRI PLUS Ultrasound, and (b) = Ultrasound. This study compares pure MRI techniques with the standard protocol.

  1. Scope: 11 countries, 25 centres;

  2. Recruitment of Participants- Criteria: eligible if (1) have not received biopsy before, (2) present clinical suspicion for prostate cancer on the basis of either abnormal results on digital rectal examination, raised PSA levels in blood (the tumour marker for prostate cancer; however, it cannot give a very accurate picture of one's cancer status), or both;

  3. Recruitment of Participants - Venue: Outpatient Clinics;

  4. Imaging Modalities: MRI (1.5, 3.0 T) vs Standard Protocol (transrectal ultrasound-guided)*;

  5. Outcomes:

(a) Patient-reported Outcomes (questions on quality of life)- immediately after intervention, 30 days after biopsy, and after MRI;


(b) Primary Outcome: Proportion of men with clinically significant prostate cancer (Gleason Score > 7);


(c) Secondary Outcomes including: Proportion of men with clinically insignificant cancer, the proportion of men in the MRI group who did not go on to have biopsy (where the accuracy of the modality determines whether one is wrongly sent for biopsy despite their having clinically insignificant disease), and proportion of men having adverse effects immediately after intervention.


Characteristics of Participants (Baseline):


There are no significant differences in terms of characteristics between the two groups; The mean age of the two groups is 64 (corrected to the nearest integer) and the differences in family history of prostate cancer and abnormal digital rectal examination are negligible.


Results:

  1. There are fewer incidents of benign tissue in the MRI group - 21 per cent compared to 40 per cent; in this case, it means that fewer patients with non-cancerous lesions are being referred for biopsy in the MRI group;

  2. 31 per cent of men in the MRI group did not go for a biopsy, significantly higher than that in the Standard Group (6 per cent);

  3. In terms of distribution of the Gleason scores of lesions, the vast majority of lesions are defined as 'clinically insignificant' in the Standard Group (22 per cent), whereas for the MRI group, clinically insignificant lesions only account for 9 per cent (this operates on the earlier definition that lesions with a Gleason score of 6 - or 3+3, are clinically insignificant);

  4. Most lesions detected by MRI + biopsied that way bear a Gleason score of 7 (3+4) - 21 per cent;

  5. In total, 95 (38 per cent) of those in the MRI group have clinically significant lesions, as compared to 64 (26 per cent) of those in the Standard Group; this accords to the results from the intention-to-treat analysis (all patients after randomisation are included, not only those who satisfied every diagnostic procedure per protocol afterwards);

  6. The patient-reported outcome profile is more favourable in the MRI group as compared to the Standard group.

From these results, the authors of the trial concluded that 'MRI with or without targeted biopsy was conclusively superior to standard transrectal ultrasonography–guided biopsy.' [2]


Before the cynics can chip in, the authors did express their concern over several issues including the 'agreement' between radiologists. When we analyse scans, it's best that we appoint two or three radiologists to do so. They should be of similar levels of experience in that particular field, as evidenced by specialism, years of experience and numbers of scans read per year. Some centres might opt for one radiologist but that's putting hefty responsibility on one's shoulders. Agreement refers to whether the same conclusions are reached between the two radiologists. The agreement rate of this study is roughly 78 per cent, which is defined as 'moderate' by the authors. I personally think 78 per cent is already an achievement, since prostate MRI has been commented as a modality requiring much practice [1]. Moreover, there is concern over the grading of the pathological lesions - however, as with all pathological assessments, there are a steep learning curve (meaning that not everyone can grasp the technique at the same rate, which is understandable if you've looked at a pathological specimen before) and comparatively underwhelming agreement rates.

Tables and figure extracted from the NEJM-published paper [2]. In Figure 2, the term 'non-inferiority margin' is used to see if two modalities are comparable. In other words, 'non-inferior', where using either does not cause detriment to the other in sole virtue of using a different modality. This is the original label used by the authors. However, as you can see, all analyses point towards the bounty of the MRI techniques that the authors reckon the MRI techniques are more superior (however, it would be nicer if the definition for superiority were stated clearly at first).

Diagram showing the Gleason's Pattern - histological differences between different grades (extracted from: https://training.seer.cancer.gov/images/prostate/gleason_pattern.jpg)

[the word 'differentiated' is use to describe the degree of aggressiveness of the tumour- the less differentiated it is, the more aggressively it develops; this is due to the retention of the stem cell-like tendency to proliferate should they be less well-differentiated, meaning that they haven't strayed too far off from their origin]

Histological Specimen of Prostate Cancer (the purple rings, in lack of a better term to describe them) (courtesy of Cleveland Clinic) [there are several characteristics to tell if cells are malignant; the easiest one is pleomorphism, meaning that the cells are diverse in size and shape; this is important since in cancers, different cells develop differently and they form 'clones'. Natural selection also occurs so that survival is reserved for the fittest, i.e. those who can proliferate the quickest and invade relentlessly]


What Next?


From the PRECISION Trial, encouraging results from the MRI group means that MRI should, at least in the ideal world, be the diagnostic modality of choice for prostate cancer, due to that more clinically-significant lesions are detected and fewer clinically-insignificant lesions are detected. This means that overtreatment can potentially be a matter of the past. However, we're better than this - science does not stop at the point where one study has identified the merits of MRI over the standard protocol.


First, we need more studies to replicate the same results. Although involving 25 centres is quite impressive and the the representativeness of the study is much higher than that of single-centred ones, more should be performed. Systematic reviews, synthesising valid clinical information from multiple trials, can therefore tell us if MRI techniques are really preferred over the standard protocol, leading to changes of current best medical practice.


Second, MRI is not just a buzzword. It encapsulates many concepts and practical issues. Chief of which, as described by Professor Mark Emberton, is the sequences used. [1] Currently, the MRI used in the PRECISION study can also be described as 'multiparametric'. Breaking down the word into key pieces, it means the utilisation of multiple sequences or parameters in determination of the presence and type of a lesion. Here, we have the following:

  1. T2 (the one which lights up when there's water - to put in a coarse but simple way);

  2. DCE - Dynamic Contrast-Enhanced (using a material which helps amplify signals);

  3. DWI - Diffusion-weighted imaging (which explores the diffusion potential of water molecules);

  4. ADC - Apparent Diffusion coefficient (similar to DWI; findings are flipped as compared to DWI).

Multiparametric MRI of Prostate Cancer (circled) - Extracted from Semantic Scholar. (I know the DCE has some cool colours, but are they really worth lowered clinical efficiency?)


What's wrong with using 4 sequences, you might ask? To commence with, there is the issue of image acquisition. When we push a patient in an MRI machine, we don't get the images straight away. Depending on the images we obtain for the same patient (dependent on the number of images per 'stack' in one modality; and the number of modalities adopted), the time of acquisition can range from fifteen to forty-five minutes. I do vividly recall that MR Brain images require around thirty minutes to be obtained. The issue with including DCE is - increased acquisition time. Current MRI acquisition time for prostate imaging is around 35 minutes, but without DCE, it'd be around 15 minutes. You might think that the difference here is minimal, but in clinical practice, if we can reduce 20 minutes of acquisition time per patient, given the scarce MRI machines available, we can scan more patients than we can now and increase the cost-effectiveness of those machines. However, removing DCE from the equation without ample justification is not going to get many compliments. We at least need to prove that DCE is negligible and radiologists can diagnose prostate cancer safely without using DCE.


PROMIS Study


A recently published paper (October 2020) explores this issue. It reports the results of the PROMIS Study (also run by UCL) looking into the question of whether DCE can be excluded from multiparametric MRI in prostate imaging. The results are rather complicated, since sets of which are synthesised according to two different definitions of the term 'clinically significant lesion'. The primary outcome utilises definition 1: Gleason score >=4 + 3 or cancer core length >=6 mm of any grade. [7]


The results are promising. I present them in the following numbered list:

  1. T2 - Sensitivity: 96%, Specificity: 30%, % with MRI score 4-5 being +ive on biopsy: 74;

  2. T2 + DWI - Sensitivity: 94%, Specificity: 37%, % with MRI score 4-5 being +ive on biopsy: 71;

  3. T2, DWI + DCE - Sensitivity: 95%, Specificity: 38%; % with MRI score 4-5 being +ive on biopsy: 69.

Note that ADC comes with DWI. You can't interpret one and not interpret the other because we need to make sure that the DWI findings are not due to artefacts (i.e. fake lesions).


Sensitivity refers to how accurate a measure is in correctly detecting a disease. Specificity refers to how accurate a measure is in correctly identifying those with no disease. As you can see, the measures cited above show that techniques not involving DCE are not significantly inferior. Moreover, more patients with MRI score 4-5, as determined by using non-DCE MRI sequences, have clinically significant lesions on biopsy. There is higher agreement between the imaging and histological modalities, meaning that MRI without DCE turns out to reflect the actual situation better. The authors thus concluded that contrast adds little when using multiparametric MRI to exclude significant prostate cancer. [7]

[7] As we can see, the general proportion of significant cancers identified (Scores 3-5) by T2 + DWI > the general proportion identified by T2 + DWI + DCE (rings true when either definition is used)


All in all, MRI is the game-changer we all need to remove the invisibility cloak of the 'malicious walnut'.


*MRI machines are graded according to T's- the strength of the magnetic field. Different centres have different protocols and financial managers so it really depends on the individual centre's resources. Some might go the distance and buy 'more powerful' machines, such as the 3.0T type. Some are less well-endowed and go for the 1.5T model. From personal experience, there is a huge difference between 1.5T and 3.0T-produced images, with the latter ones being much clearer and less prone to artefacts. This is especially true for imaging of the prostate, where there is a steep learning curve in interpretation. Although one may argue that heightened clarity may also come with the unwelcome result of showing more clinically insignificant lesions, that's not really a significant issue. Currently, in some centres such as MIT and Harvard, I've read they've got the 7.0T machine. Those images are so clear that one can even see the notorious cerebral microinfarcts in those images.


References and Further Reading:


[1] UCL Faculty of Medical Sciences. (2020, November 24). How UCL revolutionised prostate cancer diagnosis and treatment for men [Video]. Youtube. https://www.youtube.com/watch?v=XvkHfeiePPs&ab_channel=UCLFacultyofMedicalSciences.


[2] Kasivisvanathan V, Rannikko A, Borghi M, et al. (2018). MRI-Targeted or Standard Biopsy for Prostate-Cancer Diagnosis. New England Journal Of Medicine, 378(19), 1767-1777. https://doi.org/10.1056/nejmoa1801993.


[3] Jahn JL, Giovannucci EL, Stampfer MJ. (2015). The high prevalence of undiagnosed prostate cancer at autopsy: implications for epidemiology and treatment of prostate cancer in the Prostate-specific Antigen-era. International journal of cancer, 137(12), 2795–2802. https://doi.org/10.1002/ijc.29408.


[4] Liau J, Goldberg D, Arif-Tiwari H. (2019). Prostate Cancer Detection and Diagnosis: Role of Ultrasound with MRI Correlates. Current Radiology Reports, 7(3). https://doi.org/10.1007/s40134-019-0318-8.


[5] What is a Gleason Score? | Prostate Cancer Foundation. Prostate Cancer Foundation. (2020). Retrieved 5 December 2020, from https://www.pcf.org/about-prostate-cancer/diagnosis-staging-prostate-cancer/gleason-score-isup-grade/#:~:text=How%20is%20the%20Gleason%20Score,to%20determine%20your%20Gleason%20score.


[6] Thomsen H. (2011). Contrast media safety—An update. European Journal Of Radiology, 80(1), 77-82. https://doi.org/10.1016/j.ejrad.2010.12.104.


[7] Bosaily A, Frangou E, Ahmed H, et al. (2020). Additional Value of Dynamic Contrast-enhanced Sequences in Multiparametric Prostate Magnetic Resonance Imaging: Data from the PROMIS Study. European Urology, 78(4), 503-511. https://doi.org/10.1016/j.eururo.2020.03.002.


[8] Grummet J, Gorin M, Popert R, et al. (2020). “TREXIT 2020”: why the time to abandon transrectal prostate biopsy starts now. Prostate Cancer And Prostatic Diseases, 23(1), 62-65. https://doi.org/10.1038/s41391-020-0204-8.

[an opinion piece on whether the current protocol of obtaining prostate biopsies through the rectum should be abandoned in favour of the trans-perineal approach, which gains access through the perineum, the strip of skin between the base of the male genitalia and the anus]

6 views0 comments

Comments


Post: Blog2_Post
bottom of page